fix additional files note (#4699)

* fix additional files note

* Trigger CI

* Trigger CI

.clang-tidy

@@ -1,4 +1,4 @@
-Checks: 'modernize-*,-modernize-make-*,-modernize-raw-string-literal,google-*,-google-default-arguments,-clang-diagnostic-#pragma-messages,readability-identifier-naming'
+Checks: 'modernize-*,-modernize-make-*,-modernize-use-auto,-modernize-raw-string-literal,google-*,-google-default-arguments,-clang-diagnostic-#pragma-messages,readability-identifier-naming'
 CheckOptions:
   - { key: readability-identifier-naming.ClassCase,           value: CamelCase  }
   - { key: readability-identifier-naming.StructCase,          value: CamelCase  }
@@ -6,8 +6,8 @@ CheckOptions:
   - { key: readability-identifier-naming.TypedefCase,          value: CamelCase  }
   - { key: readability-identifier-naming.TypeTemplateParameterCase,          value: CamelCase  }
   - { key: readability-identifier-naming.MemberCase,          value: lower_case  }
-  - { key: readability-identifier-naming.PrivateMemberSuffix,           value: '_'  }
-  - { key: readability-identifier-naming.ProtectedMemberSuffix,           value: '_'  }
+  - { key: readability-identifier-naming.PrivateMemberSuffix,    value: '_'  }
+  - { key: readability-identifier-naming.ProtectedMemberSuffix,  value: '_'  }
   - { key: readability-identifier-naming.EnumCase,          value: CamelCase  }
   - { key: readability-identifier-naming.EnumConstant,          value: CamelCase  }
   - { key: readability-identifier-naming.EnumConstantPrefix,          value: k  }

.github/lock.yml

@@ -0,0 +1,32 @@
+# Configuration for lock-threads - https://github.com/dessant/lock-threads
+
+# Number of days of inactivity before a closed issue or pull request is locked
+daysUntilLock: 90
+
+# Issues and pull requests with these labels will not be locked. Set to `[]` to disable
+exemptLabels:
+  - feature-request
+
+# Label to add before locking, such as `outdated`. Set to `false` to disable
+lockLabel: false
+
+# Comment to post before locking. Set to `false` to disable
+lockComment: false
+
+# Assign `resolved` as the reason for locking. Set to `false` to disable
+setLockReason: true
+
+# Limit to only `issues` or `pulls`
+# only: issues
+
+# Optionally, specify configuration settings just for `issues` or `pulls`
+# issues:
+#   exemptLabels:
+#     - help-wanted
+#   lockLabel: outdated
+
+# pulls:
+#   daysUntilLock: 30
+
+# Repository to extend settings from
+# _extends: repo

.gitignore

@@ -91,3 +91,8 @@ lib/
 metastore_db
 
 plugin/updater_gpu/test/cpp/data
+/include/xgboost/build_config.h
+
+# files from R-package source install
+**/config.status
+R-package/src/Makevars

.travis.yml

@@ -3,65 +3,30 @@ sudo: required
 
 # Enabling test on Linux and OS X
 os:
-  - linux
   - osx
 
-osx_image: xcode8
-
-group: deprecated-2017Q4
+osx_image: xcode9.3
 
 # Use Build Matrix to do lint and build seperately
 env:
   matrix:
-    # code lint
-    - TASK=lint
-    # r package test
-    - TASK=r_test
     # python package test
     - TASK=python_test
-    - TASK=python_lightweight_test
     # java package test
     - TASK=java_test
     # cmake test
-    - TASK=cmake_test
-    # c++ test
-    - TASK=cpp_test
-    # distributed test
-    - TASK=distributed_test
-
-matrix:
-  exclude:
-    - os: osx
-      env: TASK=lint
-    - os: osx
-      env: TASK=cmake_test
-    - os: linux
-      env: TASK=r_test
-    - os: osx
-      env: TASK=python_lightweight_test
-    - os: osx
-      env: TASK=cpp_test
-    - os: osx
-      env: TASK=distributed_test
+    # - TASK=cmake_test
 
 # dependent apt packages
 addons:
-  apt:
-    sources:
-      - llvm-toolchain-trusty-5.0
-      - ubuntu-toolchain-r-test
-      - george-edison55-precise-backports
+  homebrew:
     packages:
-      - clang
-      - clang-tidy-5.0
-      - cmake-data
-      - doxygen
-      - wget
-      - libcurl4-openssl-dev
-      - unzip
+      - gcc@7
       - graphviz
-      - gcc-4.8
-      - g++-4.8
+      - openssl
+      - libgit2
+      - r
+    update: true
 
 before_install:
   - source dmlc-core/scripts/travis/travis_setup_env.sh

CMakeLists.txt

@@ -1,258 +1,229 @@
-cmake_minimum_required (VERSION 3.2)
-project(xgboost)
+cmake_minimum_required(VERSION 3.3)
+project(xgboost LANGUAGES CXX C VERSION 0.90)
 include(cmake/Utils.cmake)
 list(APPEND CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/cmake/modules")
-find_package(OpenMP)
+cmake_policy(SET CMP0022 NEW)
+
+message(STATUS "CMake version ${CMAKE_VERSION}")
+if (MSVC)
+  cmake_minimum_required(VERSION 3.11)
+endif (MSVC)
 
 set_default_configuration_release()
-msvc_use_static_runtime()
 
-# Options
-option(USE_CUDA  "Build with GPU acceleration")
-option(USE_AVX  "Build with AVX instructions. May not produce identical results due to approximate math." OFF)
-option(USE_NCCL "Build using NCCL for multi-GPU. Also requires USE_CUDA")
+#-- Options
+option(BUILD_C_DOC "Build documentation for C APIs using Doxygen." OFF)
+option(USE_OPENMP "Build with OpenMP support." ON)
+## Bindings
 option(JVM_BINDINGS "Build JVM bindings" OFF)
-option(GOOGLE_TEST "Build google tests" OFF)
 option(R_LIB "Build shared library for R package" OFF)
-option(USE_SANITIZER "Use santizer flags" OFF)
+## Dev
+option(GOOGLE_TEST "Build google tests" OFF)
+option(USE_DMLC_GTEST "Use google tests bundled with dmlc-core submodule (EXPERIMENTAL)" OFF)
+option(USE_NVTX "Build with cuda profiling annotations. Developers only." OFF)
+set(NVTX_HEADER_DIR "" CACHE PATH "Path to the stand-alone nvtx header")
+## CUDA
+option(USE_CUDA  "Build with GPU acceleration" OFF)
+option(USE_NCCL  "Build with NCCL to enable multi-GPU support." OFF)
+option(BUILD_WITH_SHARED_NCCL "Build with shared NCCL library." OFF)
 set(GPU_COMPUTE_VER "" CACHE STRING
-  "Space separated list of compute versions to be built against, e.g. '35 61'")
+  "Semicolon separated list of compute versions to be built against, e.g. '35;61'")
+if (BUILD_WITH_SHARED_NCCL AND (NOT USE_NCCL))
+  message(SEND_ERROR "Build XGBoost with -DUSE_NCCL=ON to enable BUILD_WITH_SHARED_NCCL.")
+endif (BUILD_WITH_SHARED_NCCL AND (NOT USE_NCCL))
+## Sanitizers
+option(USE_SANITIZER "Use santizer flags" OFF)
+option(SANITIZER_PATH "Path to sanitizes.")
 set(ENABLED_SANITIZERS "address" "leak" CACHE STRING
   "Semicolon separated list of sanitizer names. E.g 'address;leak'. Supported sanitizers are
 address, leak and thread.")
+## Plugins
+option(PLUGIN_LZ4 "Build lz4 plugin" OFF)
+option(PLUGIN_DENSE_PARSER "Build dense parser plugin" OFF)
 
-# Deprecation warning
-if(PLUGIN_UPDATER_GPU)
-  set(USE_CUDA ON)
-  message(WARNING "The option 'PLUGIN_UPDATER_GPU' is deprecated. Set 'USE_CUDA' instead.")
-endif()
-
-# Compiler flags
-set(CMAKE_CXX_STANDARD 11)
-set(CMAKE_CXX_STANDARD_REQUIRED ON)
-if(OpenMP_CXX_FOUND OR OPENMP_FOUND)
-  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${OpenMP_CXX_FLAGS}")
-endif()
-set(CMAKE_POSITION_INDEPENDENT_CODE ON)
-if(MSVC)
-  # Multithreaded compilation
-  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /MP")
-else()
-  # Correct error for GCC 5 and cuda
-  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -D_MWAITXINTRIN_H_INCLUDED -D_FORCE_INLINES")
-  # Performance
-  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -funroll-loops")
-endif()
-if(WIN32 AND MINGW)
-  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -static-libstdc++")
-endif()
+## Deprecation warning
+if (USE_AVX)
+  message(WARNING "The option 'USE_AVX' is deprecated as experimental AVX features have been removed from xgboost.")
+endif (USE_AVX)
 
 # Sanitizer
-if(USE_SANITIZER)
+if (USE_SANITIZER)
+  # Older CMake versions have had troubles with Sanitizer
+  cmake_minimum_required(VERSION 3.12)
   include(cmake/Sanitizer.cmake)
   enable_sanitizers("${ENABLED_SANITIZERS}")
-endif(USE_SANITIZER)
+endif (USE_SANITIZER)
 
-# AVX
-if(USE_AVX)
-  if(MSVC)
-    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /arch:AVX")
-  else()
-    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mavx")
-  endif()
-  add_definitions(-DXGBOOST_USE_AVX)
-endif()
+if (USE_CUDA)
+  cmake_minimum_required(VERSION 3.12)
+  SET(USE_OPENMP ON CACHE BOOL "CUDA requires OpenMP" FORCE)
+  # `export CXX=' is ignored by CMake CUDA.
+  set(CMAKE_CUDA_HOST_COMPILER ${CMAKE_CXX_COMPILER})
+  message(STATUS "Configured CUDA host compiler: ${CMAKE_CUDA_HOST_COMPILER}")
+
+  enable_language(CUDA)
+  set(GEN_CODE "")
+  format_gencode_flags("${GPU_COMPUTE_VER}" GEN_CODE)
+  message(STATUS "CUDA GEN_CODE: ${GEN_CODE}")
+endif (USE_CUDA)
 
 # dmlc-core
-add_subdirectory(dmlc-core)
-set(LINK_LIBRARIES dmlc rabit)
-
-# enable custom logging
-add_definitions(-DDMLC_LOG_CUSTOMIZE=1)
-
-# compiled code customizations for R package
-if(R_LIB)
-  add_definitions(
-    -DXGBOOST_STRICT_R_MODE=1
-    -DXGBOOST_CUSTOMIZE_GLOBAL_PRNG=1
-    -DDMLC_LOG_BEFORE_THROW=0
-    -DDMLC_DISABLE_STDIN=1
-    -DDMLC_LOG_CUSTOMIZE=1
-    -DRABIT_CUSTOMIZE_MSG_
-    -DRABIT_STRICT_CXX98_
-  )
-endif()
-
-include_directories (
-    ${PROJECT_SOURCE_DIR}/include
-    ${PROJECT_SOURCE_DIR}/dmlc-core/include
-    ${PROJECT_SOURCE_DIR}/rabit/include
-)
-
-file(GLOB_RECURSE SOURCES
-    src/*.cc
-    src/*.h
-    include/*.h
-)
-
-# Only add main function for executable target
-list(REMOVE_ITEM SOURCES ${PROJECT_SOURCE_DIR}/src/cli_main.cc)
-
-file(GLOB_RECURSE CUDA_SOURCES
-    src/*.cu
-    src/*.cuh
-)
+msvc_use_static_runtime()
+add_subdirectory(${PROJECT_SOURCE_DIR}/dmlc-core)
+set_target_properties(dmlc PROPERTIES
+  CXX_STANDARD 11
+  CXX_STANDARD_REQUIRED ON
+  POSITION_INDEPENDENT_CODE ON)
+list(APPEND LINKED_LIBRARIES_PRIVATE dmlc)
 
 # rabit
-# TODO: Create rabit cmakelists.txt
-set(RABIT_SOURCES
+# full rabit doesn't build on windows, so we can't import it as subdirectory
+if(MINGW OR R_LIB)
+  set(RABIT_SOURCES
+    rabit/src/engine_empty.cc
+    rabit/src/c_api.cc)
+else ()
+  set(RABIT_SOURCES
     rabit/src/allreduce_base.cc
     rabit/src/allreduce_robust.cc
     rabit/src/engine.cc
-    rabit/src/c_api.cc
-)
-set(RABIT_EMPTY_SOURCES
-    rabit/src/engine_empty.cc
-    rabit/src/c_api.cc
-)
-if(MINGW OR R_LIB)
-  # build a dummy rabit library
-  add_library(rabit STATIC ${RABIT_EMPTY_SOURCES})
-else()
-  add_library(rabit STATIC ${RABIT_SOURCES})
-endif()
+    rabit/src/c_api.cc)
+endif (MINGW OR R_LIB)
+add_library(rabit STATIC ${RABIT_SOURCES})
+target_include_directories(rabit PRIVATE
+  $<BUILD_INTERFACE:${CMAKE_CURRENT_LIST_DIR}/dmlc-core/include>
+  $<BUILD_INTERFACE:${CMAKE_CURRENT_LIST_DIR}/rabit/include/rabit>)
+set_target_properties(rabit
+  PROPERTIES
+  CXX_STANDARD 11
+  CXX_STANDARD_REQUIRED ON
+  POSITION_INDEPENDENT_CODE ON)
+list(APPEND LINKED_LIBRARIES_PRIVATE rabit)
 
-if(USE_CUDA)
-  find_package(CUDA 8.0 REQUIRED)
-  cmake_minimum_required(VERSION 3.5)
+# Exports some R specific definitions and objects
+if (R_LIB)
+  add_subdirectory(${PROJECT_SOURCE_DIR}/R-package)
+endif (R_LIB)
 
-  add_definitions(-DXGBOOST_USE_CUDA)
+# core xgboost
+add_subdirectory(${PROJECT_SOURCE_DIR}/src)
+set(XGBOOST_OBJ_SOURCES "${XGBOOST_OBJ_SOURCES};$<TARGET_OBJECTS:objxgboost>")
 
-  include_directories(cub)
+#-- Shared library
+add_library(xgboost SHARED ${XGBOOST_OBJ_SOURCES} ${PLUGINS_SOURCES})
+target_include_directories(xgboost
+  INTERFACE
+  $<INSTALL_INTERFACE:${CMAKE_INSTALL_PREFIX}/include>
+  $<BUILD_INTERFACE:${CMAKE_CURRENT_LIST_DIR}/include>)
+target_link_libraries(xgboost PRIVATE ${LINKED_LIBRARIES_PRIVATE})
 
-  if(USE_NCCL)
-    find_package(Nccl REQUIRED)
-    include_directories(${NCCL_INCLUDE_DIR})
-    add_definitions(-DXGBOOST_USE_NCCL)
-  endif()
+# This creates its own shared library `xgboost4j'.
+if (JVM_BINDINGS)
+  add_subdirectory(${PROJECT_SOURCE_DIR}/jvm-packages)
+endif (JVM_BINDINGS)
+#-- End shared library
 
-  set(GENCODE_FLAGS "")
-  format_gencode_flags("${GPU_COMPUTE_VER}" GENCODE_FLAGS)
-  message("cuda architecture flags: ${GENCODE_FLAGS}")
+#-- CLI for xgboost
+add_executable(runxgboost ${PROJECT_SOURCE_DIR}/src/cli_main.cc ${XGBOOST_OBJ_SOURCES})
+# For cli_main.cc only
+if (USE_OPENMP)
+  find_package(OpenMP REQUIRED)
+  target_compile_options(runxgboost PRIVATE ${OpenMP_CXX_FLAGS})
+endif (USE_OPENMP)
+target_include_directories(runxgboost
+  PRIVATE
+  ${PROJECT_SOURCE_DIR}/include
+  ${PROJECT_SOURCE_DIR}/dmlc-core/include
+  ${PROJECT_SOURCE_DIR}/rabit/include)
+target_link_libraries(runxgboost PRIVATE ${LINKED_LIBRARIES_PRIVATE})
+set_target_properties(
+  runxgboost PROPERTIES
+  OUTPUT_NAME xgboost
+  CXX_STANDARD 11
+  CXX_STANDARD_REQUIRED ON)
+#-- End CLI for xgboost
 
-  set(CUDA_NVCC_FLAGS "${CUDA_NVCC_FLAGS};--expt-extended-lambda;--expt-relaxed-constexpr;${GENCODE_FLAGS};-lineinfo;")
-  if(NOT MSVC)
-    set(CUDA_NVCC_FLAGS "${CUDA_NVCC_FLAGS};-Xcompiler -fPIC; -Xcompiler -Werror; -std=c++11")
-  endif()
+set_output_directory(runxgboost ${PROJECT_SOURCE_DIR})
+set_output_directory(xgboost ${PROJECT_SOURCE_DIR}/lib)
+# Ensure these two targets do not build simultaneously, as they produce outputs with conflicting names
+add_dependencies(xgboost runxgboost)
 
-  cuda_add_library(gpuxgboost ${CUDA_SOURCES} STATIC)
-
-  if(USE_NCCL)
-    link_directories(${NCCL_LIBRARY})
-    target_link_libraries(gpuxgboost ${NCCL_LIB_NAME})
-  endif()
-  list(APPEND LINK_LIBRARIES gpuxgboost)
-endif()
-
-
-# flags and sources for R-package
-if(R_LIB)
-  file(GLOB_RECURSE R_SOURCES
-    R-package/src/*.h
-    R-package/src/*.c
-    R-package/src/*.cc
-  )
-  list(APPEND SOURCES ${R_SOURCES})
-endif()
-
-add_library(objxgboost OBJECT ${SOURCES})
-
-
-# building shared library for R package
-if(R_LIB)
-  find_package(LibR REQUIRED)
-
-  list(APPEND LINK_LIBRARIES "${LIBR_CORE_LIBRARY}")
-  MESSAGE(STATUS "LIBR_CORE_LIBRARY " ${LIBR_CORE_LIBRARY})
-
-  include_directories(
-    "${LIBR_INCLUDE_DIRS}"
-    "${PROJECT_SOURCE_DIR}"
-  )
-
-  # Shared library target for the R package
-  add_library(xgboost SHARED $<TARGET_OBJECTS:objxgboost>)
-  target_link_libraries(xgboost ${LINK_LIBRARIES})
-  # R uses no lib prefix in shared library names of its packages
+#-- Installing XGBoost
+if (R_LIB)
   set_target_properties(xgboost PROPERTIES PREFIX "")
-
+  if (APPLE)
+    set_target_properties(xgboost PROPERTIES SUFFIX ".so")
+  endif (APPLE)
   setup_rpackage_install_target(xgboost ${CMAKE_CURRENT_BINARY_DIR})
-  # use a dummy location for any other remaining installs
   set(CMAKE_INSTALL_PREFIX "${CMAKE_CURRENT_BINARY_DIR}/dummy_inst")
+endif (R_LIB)
+if (MINGW)
+  set_target_properties(xgboost PROPERTIES PREFIX "")
+endif (MINGW)
 
-# main targets: shared library & exe
-else()
-  # Executable
-  add_executable(runxgboost $<TARGET_OBJECTS:objxgboost> src/cli_main.cc)
-  set_target_properties(runxgboost PROPERTIES
-    OUTPUT_NAME xgboost
-  )
-  set_output_directory(runxgboost ${PROJECT_SOURCE_DIR})
-  target_link_libraries(runxgboost ${LINK_LIBRARIES})
+if (BUILD_C_DOC)
+  include(cmake/Doc.cmake)
+  run_doxygen()
+endif (BUILD_C_DOC)
 
-  # Shared library
-  add_library(xgboost SHARED $<TARGET_OBJECTS:objxgboost>)
-  target_link_libraries(xgboost ${LINK_LIBRARIES})
-  set_output_directory(xgboost ${PROJECT_SOURCE_DIR}/lib)
-  if(MINGW)
-    # remove the 'lib' prefix to conform to windows convention for shared library names
-    set_target_properties(xgboost PROPERTIES PREFIX "")
-  endif()
+include(GNUInstallDirs)
+# Exposing only C APIs.
+install(FILES
+  "${PROJECT_SOURCE_DIR}/include/xgboost/c_api.h"
+  DESTINATION
+  include/xgboost/)
 
-  #Ensure these two targets do not build simultaneously, as they produce outputs with conflicting names
-  add_dependencies(xgboost runxgboost)
-endif()
+install(TARGETS xgboost runxgboost
+  EXPORT XGBoostTargets
+  ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR}
+  LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
+  RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR}
+  INCLUDES DESTINATION ${LIBLEGACY_INCLUDE_DIRS})
+install(EXPORT XGBoostTargets
+  FILE XGBoostTargets.cmake
+  NAMESPACE xgboost::
+  DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/xgboost)
 
+include(CMakePackageConfigHelpers)
+configure_package_config_file(
+  ${CMAKE_CURRENT_LIST_DIR}/cmake/xgboost-config.cmake.in
+  ${CMAKE_CURRENT_BINARY_DIR}/cmake/xgboost-config.cmake
+  INSTALL_DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/xgboost)
+write_basic_package_version_file(
+  ${CMAKE_BINARY_DIR}/cmake/xgboost-config-version.cmake
+  VERSION ${XGBOOST_VERSION}
+  COMPATIBILITY AnyNewerVersion)
+install(
+  FILES
+  ${CMAKE_BINARY_DIR}/cmake/xgboost-config.cmake
+  ${CMAKE_BINARY_DIR}/cmake/xgboost-config-version.cmake
+  DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/xgboost)
 
-# JVM
-if(JVM_BINDINGS)
-    find_package(JNI QUIET REQUIRED)
-
-    include_directories(${JNI_INCLUDE_DIRS} jvm-packages/xgboost4j/src/native)
-
-    add_library(xgboost4j SHARED
-        $<TARGET_OBJECTS:objxgboost>
-        jvm-packages/xgboost4j/src/native/xgboost4j.cpp)
-    set_output_directory(xgboost4j ${PROJECT_SOURCE_DIR}/lib)
-    target_link_libraries(xgboost4j
-        ${LINK_LIBRARIES}
-        ${JAVA_JVM_LIBRARY})
-endif()
-
-
-# Test
-if(GOOGLE_TEST)
+#-- Test
+if (GOOGLE_TEST)
   enable_testing()
-  find_package(GTest REQUIRED)
+  # Unittests.
+  add_subdirectory(${PROJECT_SOURCE_DIR}/tests/cpp)
+  add_test(
+    NAME TestXGBoostLib
+    COMMAND testxgboost
+    WORKING_DIRECTORY ${PROJECT_BINARY_DIR})
 
-  file(GLOB_RECURSE TEST_SOURCES "tests/cpp/*.cc")
-  auto_source_group("${TEST_SOURCES}")
-  include_directories(${GTEST_INCLUDE_DIRS})
+  # CLI tests
+  configure_file(
+    ${PROJECT_SOURCE_DIR}/tests/cli/machine.conf.in
+    ${PROJECT_BINARY_DIR}/tests/cli/machine.conf
+    @ONLY)
+  add_test(
+    NAME TestXGBoostCLI
+    COMMAND runxgboost ${PROJECT_BINARY_DIR}/tests/cli/machine.conf
+    WORKING_DIRECTORY ${PROJECT_BINARY_DIR})
+  set_tests_properties(TestXGBoostCLI
+    PROPERTIES
+    PASS_REGULAR_EXPRESSION ".*test-rmse:0.087.*")
+endif (GOOGLE_TEST)
 
-  if(USE_CUDA)
-    file(GLOB_RECURSE CUDA_TEST_SOURCES "tests/cpp/*.cu")
-    cuda_compile(CUDA_TEST_OBJS ${CUDA_TEST_SOURCES})
-  else()
-    set(CUDA_TEST_OBJS "")
-  endif()
-
-  add_executable(testxgboost ${TEST_SOURCES} ${CUDA_TEST_OBJS} $<TARGET_OBJECTS:objxgboost>)
-  set_output_directory(testxgboost ${PROJECT_SOURCE_DIR})
-  target_link_libraries(testxgboost ${GTEST_LIBRARIES} ${LINK_LIBRARIES})
-
-  add_test(TestXGBoost testxgboost)
-endif()
-
-
-# Group sources
-auto_source_group("${SOURCES}")
+# For MSVC: Call msvc_use_static_runtime() once again to completely
+# replace /MD with /MT. See https://github.com/dmlc/xgboost/issues/4462
+# for issues caused by mixing of /MD and /MT flags
+msvc_use_static_runtime()

CONTRIBUTORS.md

@@ -6,21 +6,30 @@ Committers
 ----------
 Committers are people who have made substantial contribution to the project and granted write access to the project.
 * [Tianqi Chen](https://github.com/tqchen), University of Washington
-  - Tianqi is a PhD working on large-scale machine learning, he is the creator of the project.
+  - Tianqi is a Ph.D. student working on large-scale machine learning. He is the creator of the project.
 * [Tong He](https://github.com/hetong007), Amazon AI
-  - Tong is an applied scientist in Amazon AI, he is the maintainer of xgboost R package.
+  - Tong is an applied scientist in Amazon AI. He is the maintainer of XGBoost R package.
 * [Vadim Khotilovich](https://github.com/khotilov)
   - Vadim contributes many improvements in R and core packages.
 * [Bing Xu](https://github.com/antinucleon)
-  - Bing is the original creator of xgboost python package and currently the maintainer of [XGBoost.jl](https://github.com/antinucleon/XGBoost.jl).
+  - Bing is the original creator of XGBoost Python package and currently the maintainer of [XGBoost.jl](https://github.com/antinucleon/XGBoost.jl).
 * [Michael Benesty](https://github.com/pommedeterresautee)
-  - Micheal is a lawyer, data scientist in France, he is the creator of xgboost interactive analysis module in R.
-* [Yuan Tang](https://github.com/terrytangyuan)
-  - Yuan is a data scientist in Chicago, US. He contributed mostly in R and Python packages.
-* [Nan Zhu](https://github.com/CodingCat)
-  - Nan is a software engineer in Microsoft. He contributed mostly in JVM packages.
-* [Sergei Lebedev](https://github.com/superbobry)
-  - Serget is a software engineer in Criteo. He contributed mostly in JVM packages.
+  - Michael is a lawyer and data scientist in France. He is the creator of XGBoost interactive analysis module in R.
+* [Yuan Tang](https://github.com/terrytangyuan), Ant Financial
+  - Yuan is a software engineer in Ant Financial. He contributed mostly in R and Python packages.
+* [Nan Zhu](https://github.com/CodingCat), Uber
+  - Nan is a software engineer in Uber. He contributed mostly in JVM packages.
+* [Sergei Lebedev](https://github.com/superbobry), Criteo
+  - Sergei is a software engineer in Criteo. He contributed mostly in JVM packages.
+* [Hongliang Liu](https://github.com/phunterlau)
+* [Scott Lundberg](http://scottlundberg.com/), University of Washington
+  - Scott is a Ph.D. student at University of Washington. He is the creator of SHAP, a unified approach to explain the output of machine learning models such as decision tree ensembles. He also helps maintain the XGBoost Julia package.
+* [Rory Mitchell](https://github.com/RAMitchell), University of Waikato
+  - Rory is a Ph.D. student at University of Waikato. He is the original creator of the GPU training algorithms. He improved the CMake build system and continuous integration. 
+* [Hyunsu Cho](http://hyunsu-cho.io/), Amazon AI
+  - Hyunsu is an applied scientist in Amazon AI. He is the maintainer of the XGBoost Python package. He also manages the Jenkins continuous integration system (https://xgboost-ci.net/). He is the initial author of the CPU 'hist' updater. 
+* [Jiaming](https://github.com/trivialfis)
+  - Jiaming contributed to the GPU algorithms. He has also introduced new abstractions to improve the quality of the C++ codebase.
 
 Become a Committer
 ------------------
@@ -36,28 +45,25 @@ List of Contributors
 * [Full List of Contributors](https://github.com/dmlc/xgboost/graphs/contributors)
   - To contributors: please add your name to the list when you submit a patch to the project:)
 * [Kailong Chen](https://github.com/kalenhaha)
-  - Kailong is an early contributor of xgboost, he is creator of ranking objectives in xgboost.
+  - Kailong is an early contributor of XGBoost, he is creator of ranking objectives in XGBoost.
 * [Skipper Seabold](https://github.com/jseabold)
-  - Skipper is the major contributor to the scikit-learn module of xgboost.
+  - Skipper is the major contributor to the scikit-learn module of XGBoost.
 * [Zygmunt Zając](https://github.com/zygmuntz)
   - Zygmunt is the master behind the early stopping feature frequently used by kagglers.
 * [Ajinkya Kale](https://github.com/ajkl)
 * [Boliang Chen](https://github.com/cblsjtu)
 * [Yangqing Men](https://github.com/yanqingmen)
-  - Yangqing is the creator of xgboost java package.
+  - Yangqing is the creator of XGBoost java package.
 * [Engpeng Yao](https://github.com/yepyao)
 * [Giulio](https://github.com/giuliohome)
-  - Giulio is the creator of windows project of xgboost
+  - Giulio is the creator of Windows project of XGBoost
 * [Jamie Hall](https://github.com/nerdcha)
-  - Jamie is the initial creator of xgboost sklearn module.
+  - Jamie is the initial creator of XGBoost scikit-learn module.
 * [Yen-Ying Lee](https://github.com/white1033)
 * [Masaaki Horikoshi](https://github.com/sinhrks)
-  - Masaaki is the initial creator of xgboost python plotting module.
-* [Hongliang Liu](https://github.com/phunterlau)
-* [Hyunsu Cho](http://hyunsu-cho.io/)
-  - Hyunsu is the maintainer of the XGBoost Python package. He is in charge of submitting the Python package to Python Package Index (PyPI). He is also the initial author of the CPU 'hist' updater.
+  - Masaaki is the initial creator of XGBoost Python plotting module.
 * [daiyl0320](https://github.com/daiyl0320)
-  - daiyl0320 contributed patch to xgboost distributed version more robust, and scales stably on TB scale datasets.
+  - daiyl0320 contributed patch to XGBoost distributed version more robust, and scales stably on TB scale datasets.
 * [Huayi Zhang](https://github.com/irachex)
 * [Johan Manders](https://github.com/johanmanders)
 * [yoori](https://github.com/yoori)
@@ -68,8 +74,6 @@ List of Contributors
 * [Alex Bain](https://github.com/convexquad)
 * [Baltazar Bieniek](https://github.com/bbieniek)
 * [Adam Pocock](https://github.com/Craigacp)
-* [Rory Mitchell](https://github.com/RAMitchell)
-  - Rory is the author of the GPU plugin and also contributed the cmake build system and windows continuous integration
 * [Gideon Whitehead](https://github.com/gaw89)
 * [Yi-Lin Juang](https://github.com/frankyjuang)
 * [Andrew Hannigan](https://github.com/andrewhannigan)
@@ -78,3 +82,10 @@ List of Contributors
 * [Pierre de Sahb](https://github.com/pdesahb)
 * [liuliang01](https://github.com/liuliang01)
   - liuliang01 added support for the qid column for LibSVM input format. This makes ranking task easier in distributed setting.
+* [Andrew Thia](https://github.com/BlueTea88)
+  - Andrew Thia implemented feature interaction constraints
+* [Wei Tian](https://github.com/weitian)
+* [Chen Qin](https://github.com/chenqin)
+* [Sam Wilkinson](https://samwilkinson.io)
+* [Matthew Jones](https://github.com/mt-jones)
+* [Jiaxiang Li](https://github.com/JiaxiangBU)

Jenkinsfile

@@ -6,70 +6,93 @@
 // Command to run command inside a docker container
 dockerRun = 'tests/ci_build/ci_build.sh'
 
-def buildMatrix = [
-    [ "enabled": true,  "os" : "linux", "withGpu": true, "withNccl": true,  "withOmp": true, "pythonVersion": "2.7", "cudaVersion": "9.2" ],
-    [ "enabled": true,  "os" : "linux", "withGpu": true, "withNccl": true,  "withOmp": true, "pythonVersion": "2.7", "cudaVersion": "8.0" ],
-    [ "enabled": true,  "os" : "linux", "withGpu": true, "withNccl": false, "withOmp": true, "pythonVersion": "2.7", "cudaVersion": "8.0" ],
-]
-
 pipeline {
-    // Each stage specify its own agent
-    agent none
+  // Each stage specify its own agent
+  agent none
 
-    // Setup common job properties
-    options {
-        ansiColor('xterm')
-        timestamps()
-        timeout(time: 120, unit: 'MINUTES')
-        buildDiscarder(logRotator(numToKeepStr: '10'))
-    }
+  environment {
+    DOCKER_CACHE_REPO = '492475357299.dkr.ecr.us-west-2.amazonaws.com'
+  }
 
-    // Build stages
-    stages {
-        stage('Get sources') {
-            agent any
-            steps {
-                checkoutSrcs()
-                stash name: 'srcs', excludes: '.git/'
-                milestone label: 'Sources ready', ordinal: 1
-            }
-        }
-        stage('Build doc') {
-            agent any
-            steps {
-                script {
-                    if (env.CHANGE_ID == null) {  // This is a branch
-                        def commit_id = "${GIT_COMMIT}"
-                        def branch_name = "${GIT_LOCAL_BRANCH}"
-                        echo 'Building doc...'
-                        dir ('jvm-packages') {
-                            sh "bash ./build_doc.sh ${commit_id}"
-                            archiveArtifacts artifacts: "${commit_id}.tar.bz2", allowEmptyArchive: true
-                            echo 'Deploying doc...'
-                            withAWS(credentials:'xgboost-doc-bucket') {
-                                s3Upload file: "${commit_id}.tar.bz2", bucket: 'xgboost-docs', acl: 'PublicRead', path: "${branch_name}.tar.bz2"
-                            }
-                        }
-                    } else {                      // This is a pull request
-                        echo 'Skipping doc build step for pull request'
-                    }
-                }
-            }
-        }
-        stage('Build & Test') {
-            steps {
-                script {
-                    parallel (buildMatrix.findAll{it['enabled']}.collectEntries{ c ->
-                        def buildName = getBuildName(c)
-                        buildFactory(buildName, c)
-                    })
-                }
-            }
+  // Setup common job properties
+  options {
+    ansiColor('xterm')
+    timestamps()
+    timeout(time: 120, unit: 'MINUTES')
+    buildDiscarder(logRotator(numToKeepStr: '10'))
+    preserveStashes()
+  }
+
+  // Build stages
+  stages {
+    stage('Jenkins Linux: Get sources') {
+      agent { label 'linux && cpu' }
+      steps {
+        script {
+          checkoutSrcs()
         }
+        stash name: 'srcs'
+        milestone ordinal: 1
+      }
     }
+    stage('Jenkins Linux: Formatting Check') {
+      agent none
+      steps {
+        script {
+          parallel ([
+            'clang-tidy': { ClangTidy() },
+            'lint': { Lint() },
+            'sphinx-doc': { SphinxDoc() },
+            'doxygen': { Doxygen() }
+          ])
+        }
+        milestone ordinal: 2
+      }
+    }
+    stage('Jenkins Linux: Build') {
+      agent none
+      steps {
+        script {
+          parallel ([
+            'build-cpu': { BuildCPU() },
+            'build-gpu-cuda8.0': { BuildCUDA(cuda_version: '8.0') },
+            'build-gpu-cuda9.0': { BuildCUDA(cuda_version: '9.0') },
+            'build-gpu-cuda10.0': { BuildCUDA(cuda_version: '10.0') },
+            'build-gpu-cuda10.1': { BuildCUDA(cuda_version: '10.1') },
+            'build-jvm-packages': { BuildJVMPackages(spark_version: '2.4.3') },
+            'build-jvm-doc': { BuildJVMDoc() }
+          ])
+        }
+        milestone ordinal: 3
+      }
+    }
+    stage('Jenkins Linux: Test') {
+      agent none
+      steps {
+        script {
+          parallel ([
+            'test-python-cpu': { TestPythonCPU() },
+            'test-python-gpu-cuda8.0': { TestPythonGPU(cuda_version: '8.0') },
+            'test-python-gpu-cuda9.0': { TestPythonGPU(cuda_version: '9.0') },
+            'test-python-gpu-cuda10.0': { TestPythonGPU(cuda_version: '10.0') },
+            'test-python-gpu-cuda10.1': { TestPythonGPU(cuda_version: '10.1') },
+            'test-python-mgpu-cuda10.1': { TestPythonGPU(cuda_version: '10.1', multi_gpu: true) },
+            'test-cpp-gpu': { TestCppGPU(cuda_version: '10.1') },
+            'test-cpp-mgpu': { TestCppGPU(cuda_version: '10.1', multi_gpu: true) },
+            'test-jvm-jdk8': { CrossTestJVMwithJDK(jdk_version: '8', spark_version: '2.4.3') },
+            'test-jvm-jdk11': { CrossTestJVMwithJDK(jdk_version: '11') },
+            'test-jvm-jdk12': { CrossTestJVMwithJDK(jdk_version: '12') },
+            'test-r-3.4.4': { TestR(use_r35: false) },
+            'test-r-3.5.3': { TestR(use_r35: true) }
+          ])
+        }
+        milestone ordinal: 4
+      }
+    }
+  }
 }
 
-// initialize source codes
+// check out source code from git
 def checkoutSrcs() {
   retry(5) {
     try {
@@ -84,68 +107,239 @@ def checkoutSrcs() {
   }
 }
 
-/**
- * Creates cmake and make builds
- */
-def buildFactory(buildName, conf) {
-    def os = conf["os"]
-    def nodeReq = conf["withGpu"] ? "${os} && gpu" : "${os}"
-    def dockerTarget = conf["withGpu"] ? "gpu" : "cpu"
-    [ ("${buildName}") : { buildPlatformCmake("${buildName}", conf, nodeReq, dockerTarget) }
-    ]
+def ClangTidy() {
+  node('linux && cpu') {
+    unstash name: 'srcs'
+    echo "Running clang-tidy job..."
+    def container_type = "clang_tidy"
+    def docker_binary = "docker"
+    def dockerArgs = "--build-arg CUDA_VERSION=9.2"
+    sh """
+    ${dockerRun} ${container_type} ${docker_binary} ${dockerArgs} tests/ci_build/clang_tidy.sh
+    """
+    deleteDir()
+  }
 }
 
-/**
- * Build platform and test it via cmake.
- */
-def buildPlatformCmake(buildName, conf, nodeReq, dockerTarget) {
-    def opts = cmakeOptions(conf)
-    // Destination dir for artifacts
-    def distDir = "dist/${buildName}"
-    def dockerArgs = ""
-    if(conf["withGpu"]){
-        dockerArgs = "--build-arg CUDA_VERSION=" + conf["cudaVersion"]
+def Lint() {
+  node('linux && cpu') {
+    unstash name: 'srcs'
+    echo "Running lint..."
+    def container_type = "cpu"
+    def docker_binary = "docker"
+    sh """
+    ${dockerRun} ${container_type} ${docker_binary} make lint
+    """
+    deleteDir()
+  }
+}
+
+def SphinxDoc() {
+  node('linux && cpu') {
+    unstash name: 'srcs'
+    echo "Running sphinx-doc..."
+    def container_type = "cpu"
+    def docker_binary = "docker"
+    def docker_extra_params = "CI_DOCKER_EXTRA_PARAMS_INIT='-e SPHINX_GIT_BRANCH=${BRANCH_NAME}'"
+    sh """#!/bin/bash
+    ${docker_extra_params} ${dockerRun} ${container_type} ${docker_binary} make -C doc html
+    """
+    deleteDir()
+  }
+}
+
+def Doxygen() {
+  node('linux && cpu') {
+    unstash name: 'srcs'
+    echo "Running doxygen..."
+    def container_type = "cpu"
+    def docker_binary = "docker"
+    sh """
+    ${dockerRun} ${container_type} ${docker_binary} tests/ci_build/doxygen.sh ${BRANCH_NAME}
+    """
+    archiveArtifacts artifacts: "build/${BRANCH_NAME}.tar.bz2", allowEmptyArchive: true
+    echo 'Uploading doc...'
+    s3Upload file: "build/${BRANCH_NAME}.tar.bz2", bucket: 'xgboost-docs', acl: 'PublicRead', path: "doxygen/${BRANCH_NAME}.tar.bz2"
+    deleteDir()
+  }
+}
+
+def BuildCPU() {
+  node('linux && cpu') {
+    unstash name: 'srcs'
+    echo "Build CPU"
+    def container_type = "cpu"
+    def docker_binary = "docker"
+    sh """
+    ${dockerRun} ${container_type} ${docker_binary} tests/ci_build/build_via_cmake.sh
+    ${dockerRun} ${container_type} ${docker_binary} build/testxgboost
+    """
+    // Sanitizer test
+    def docker_extra_params = "CI_DOCKER_EXTRA_PARAMS_INIT='-e ASAN_SYMBOLIZER_PATH=/usr/bin/llvm-symbolizer -e ASAN_OPTIONS=symbolize=1 --cap-add SYS_PTRACE'"
+    def docker_args = "--build-arg CMAKE_VERSION=3.12"
+    sh """
+    ${dockerRun} ${container_type} ${docker_binary} ${docker_args} tests/ci_build/build_via_cmake.sh -DUSE_SANITIZER=ON -DENABLED_SANITIZERS="address" \
+      -DCMAKE_BUILD_TYPE=Debug -DSANITIZER_PATH=/usr/lib/x86_64-linux-gnu/
+    ${docker_extra_params} ${dockerRun} ${container_type} ${docker_binary} build/testxgboost
+    """
+    deleteDir()
+  }
+}
+
+def BuildCUDA(args) {
+  node('linux && cpu') {
+    unstash name: 'srcs'
+    echo "Build with CUDA ${args.cuda_version}"
+    def container_type = "gpu_build"
+    def docker_binary = "docker"
+    def docker_args = "--build-arg CUDA_VERSION=${args.cuda_version}"
+    sh """
+    ${dockerRun} ${container_type} ${docker_binary} ${docker_args} tests/ci_build/build_via_cmake.sh -DUSE_CUDA=ON -DUSE_NCCL=ON -DOPEN_MP:BOOL=ON
+    ${dockerRun} ${container_type} ${docker_binary} ${docker_args} bash -c "cd python-package && rm -rf dist/* && python setup.py bdist_wheel --universal"
+    """
+    // Stash wheel for CUDA 8.0 / 9.0 target
+    if (args.cuda_version == '8.0') {
+      echo 'Stashing Python wheel...'
+      stash name: 'xgboost_whl_cuda8', includes: 'python-package/dist/*.whl'
+    } else if (args.cuda_version == '9.0') {
+      echo 'Stashing Python wheel...'
+      stash name: 'xgboost_whl_cuda9', includes: 'python-package/dist/*.whl'
+      archiveArtifacts artifacts: "python-package/dist/*.whl", allowEmptyArchive: true
+      echo 'Stashing C++ test executable (testxgboost)...'
+      stash name: 'xgboost_cpp_tests', includes: 'build/testxgboost'
     }
-    // Build node - this is returned result
-    node(nodeReq) {
-        unstash name: 'srcs'
-        echo """
-        |===== XGBoost CMake build =====
-        |  dockerTarget: ${dockerTarget}
-        |  cmakeOpts   : ${opts}
-        |=========================
-        """.stripMargin('|')
-        // Invoke command inside docker
-        sh """
-        ${dockerRun} ${dockerTarget} ${dockerArgs} tests/ci_build/build_via_cmake.sh ${opts}
-        ${dockerRun} ${dockerTarget} ${dockerArgs} tests/ci_build/test_${dockerTarget}.sh
-        ${dockerRun} ${dockerTarget} ${dockerArgs} bash -c "cd python-package; rm -f dist/*; python setup.py bdist_wheel --universal"
-        rm -rf "${distDir}"; mkdir -p "${distDir}/py"
-        cp xgboost "${distDir}"
-        cp -r lib "${distDir}"
-        cp -r python-package/dist "${distDir}/py"
-        # Test the wheel for compatibility on a barebones CPU container
-        ${dockerRun} release ${dockerArgs} bash -c " \
-            auditwheel show xgboost-*-py2-none-any.whl
-            pip install --user python-package/dist/xgboost-*-none-any.whl && \
-            python -m nose tests/python"
-        """
-        archiveArtifacts artifacts: "${distDir}/**/*.*", allowEmptyArchive: true
+    deleteDir()
+  }
+}
+
+def BuildJVMPackages(args) {
+  node('linux && cpu') {
+    unstash name: 'srcs'
+    echo "Build XGBoost4J-Spark with Spark ${args.spark_version}"
+    def container_type = "jvm"
+    def docker_binary = "docker"
+    // Use only 4 CPU cores
+    def docker_extra_params = "CI_DOCKER_EXTRA_PARAMS_INIT='--cpuset-cpus 0-3'"
+    sh """
+    ${docker_extra_params} ${dockerRun} ${container_type} ${docker_binary} tests/ci_build/build_jvm_packages.sh ${args.spark_version}
+    """
+    echo 'Stashing XGBoost4J JAR...'
+    stash name: 'xgboost4j_jar', includes: 'jvm-packages/xgboost4j/target/*.jar,jvm-packages/xgboost4j-spark/target/*.jar,jvm-packages/xgboost4j-example/target/*.jar'
+    deleteDir()
+  }
+}
+
+def BuildJVMDoc() {
+  node('linux && cpu') {
+    unstash name: 'srcs'
+    echo "Building JVM doc..."
+    def container_type = "jvm"
+    def docker_binary = "docker"
+    sh """
+    ${dockerRun} ${container_type} ${docker_binary} tests/ci_build/build_jvm_doc.sh ${BRANCH_NAME}
+    """
+    archiveArtifacts artifacts: "jvm-packages/${BRANCH_NAME}.tar.bz2", allowEmptyArchive: true
+    echo 'Uploading doc...'
+    s3Upload file: "jvm-packages/${BRANCH_NAME}.tar.bz2", bucket: 'xgboost-docs', acl: 'PublicRead', path: "${BRANCH_NAME}.tar.bz2"
+    deleteDir()
+  }
+}
+
+def TestPythonCPU() {
+  node('linux && cpu') {
+    unstash name: 'xgboost_whl_cuda9'
+    unstash name: 'srcs'
+    echo "Test Python CPU"
+    def container_type = "cpu"
+    def docker_binary = "docker"
+    sh """
+    ${dockerRun} ${container_type} ${docker_binary} tests/ci_build/test_python.sh cpu
+    """
+    deleteDir()
+  }
+}
+
+def TestPythonGPU(args) {
+  nodeReq = (args.multi_gpu) ? 'linux && mgpu' : 'linux && gpu'
+  node(nodeReq) {
+    if (args.cuda_version == '8.0') {
+      unstash name: 'xgboost_whl_cuda8'
+    } else {
+      unstash name: 'xgboost_whl_cuda9'
     }
+    unstash name: 'srcs'
+    echo "Test Python GPU: CUDA ${args.cuda_version}"
+    def container_type = "gpu"
+    def docker_binary = "nvidia-docker"
+    def docker_args = "--build-arg CUDA_VERSION=${args.cuda_version}"
+    if (args.multi_gpu) {
+      echo "Using multiple GPUs"
+      sh """
+      ${dockerRun} ${container_type} ${docker_binary} ${docker_args} tests/ci_build/test_python.sh mgpu
+      """
+    } else {
+      echo "Using a single GPU"
+      sh """
+      ${dockerRun} ${container_type} ${docker_binary} ${docker_args} tests/ci_build/test_python.sh gpu
+      """
+    }
+    deleteDir()
+  }
 }
 
-def cmakeOptions(conf) {
-    return ([
-        conf["withGpu"] ? '-DUSE_CUDA=ON' : '-DUSE_CUDA=OFF',
-        conf["withNccl"] ? '-DUSE_NCCL=ON' : '-DUSE_NCCL=OFF',
-        conf["withOmp"] ? '-DOPEN_MP:BOOL=ON' : '']
-        ).join(" ")
+def TestCppGPU(args) {
+  nodeReq = (args.multi_gpu) ? 'linux && mgpu' : 'linux && gpu'
+  node(nodeReq) {
+    unstash name: 'xgboost_cpp_tests'
+    unstash name: 'srcs'
+    echo "Test C++, CUDA ${args.cuda_version}"
+    def container_type = "gpu"
+    def docker_binary = "nvidia-docker"
+    def docker_args = "--build-arg CUDA_VERSION=${args.cuda_version}"
+    if (args.multi_gpu) {
+      echo "Using multiple GPUs"
+      sh "${dockerRun} ${container_type} ${docker_binary} ${docker_args} build/testxgboost --gtest_filter=*.MGPU_*"
+    } else {
+      echo "Using a single GPU"
+      sh "${dockerRun} ${container_type} ${docker_binary} ${docker_args} build/testxgboost --gtest_filter=-*.MGPU_*"
+    }
+    deleteDir()
+  }
 }
 
-def getBuildName(conf) {
-    def gpuLabel = conf['withGpu'] ? ("_cuda" + conf['cudaVersion'] + (conf['withNccl'] ? "_nccl" : "_nonccl")) : "_cpu"
-    def ompLabel = conf['withOmp'] ? "_omp" : ""
-    def pyLabel = "_py${conf['pythonVersion']}"
-    return "${conf['os']}${gpuLabel}${ompLabel}${pyLabel}"
+def CrossTestJVMwithJDK(args) {
+  node('linux && cpu') {
+    unstash name: 'xgboost4j_jar'
+    unstash name: 'srcs'
+    if (args.spark_version != null) {
+      echo "Test XGBoost4J on a machine with JDK ${args.jdk_version}, Spark ${args.spark_version}"
+    } else {
+      echo "Test XGBoost4J on a machine with JDK ${args.jdk_version}"
+    }
+    def container_type = "jvm_cross"
+    def docker_binary = "docker"
+    def spark_arg = (args.spark_version != null) ? "--build-arg SPARK_VERSION=${args.spark_version}" : ""
+    def docker_args = "--build-arg JDK_VERSION=${args.jdk_version} ${spark_arg}"
+    // Run integration tests only when spark_version is given
+    def docker_extra_params = (args.spark_version != null) ? "CI_DOCKER_EXTRA_PARAMS_INIT='-e RUN_INTEGRATION_TEST=1'" : ""
+    sh """
+    ${docker_extra_params} ${dockerRun} ${container_type} ${docker_binary} ${docker_args} tests/ci_build/test_jvm_cross.sh
+    """
+    deleteDir()
+  }
 }
 
+def TestR(args) {
+  node('linux && cpu') {
+    unstash name: 'srcs'
+    echo "Test R package"
+    def container_type = "rproject"
+    def docker_binary = "docker"
+    def use_r35_flag = (args.use_r35) ? "1" : "0"
+    def docker_args = "--build-arg USE_R35=${use_r35_flag}"
+    sh """
+    ${dockerRun} ${container_type} ${docker_binary} ${docker_args} tests/ci_build/build_test_rpkg.sh
+    """
+    deleteDir()
+  }
+}

Jenkinsfile-win64

@@ -0,0 +1,134 @@
+#!/usr/bin/groovy
+// -*- mode: groovy -*-
+
+/* Jenkins pipeline for Windows AMD64 target */
+
+pipeline {
+  agent none
+  // Build stages
+  stages {
+    stage('Jenkins Win64: Get sources') {
+      agent { label 'win64 && build' }
+      steps {
+        script {
+          checkoutSrcs()
+        }
+        stash name: 'srcs'
+        milestone ordinal: 1
+      }
+    }
+    stage('Jenkins Win64: Build') {
+      agent none
+      steps {
+        script {
+          parallel ([
+            'build-win64-cuda9.0': { BuildWin64() }
+          ])
+        }
+        milestone ordinal: 2
+      }
+    }
+    stage('Jenkins Win64: Test') {
+      agent none
+      steps {
+        script {
+          parallel ([
+            'test-win64-cpu': { TestWin64CPU() },
+            'test-win64-gpu-cuda9.0': { TestWin64GPU(cuda_target: 'cuda9') },
+            'test-win64-gpu-cuda10.0': { TestWin64GPU(cuda_target: 'cuda10_0') },
+            'test-win64-gpu-cuda10.1': { TestWin64GPU(cuda_target: 'cuda10_1') }
+          ])
+        }
+        milestone ordinal: 3
+      }
+    }
+  }
+}
+
+// check out source code from git
+def checkoutSrcs() {
+  retry(5) {
+    try {
+      timeout(time: 2, unit: 'MINUTES') {
+        checkout scm
+        sh 'git submodule update --init'
+      }
+    } catch (exc) {
+      deleteDir()
+      error "Failed to fetch source codes"
+    }
+  }
+}
+
+def BuildWin64() {
+  node('win64 && build') {
+    unstash name: 'srcs'
+    echo "Building XGBoost for Windows AMD64 target..."
+    bat "nvcc --version"
+    bat """
+    mkdir build
+    cd build
+    cmake .. -G"Visual Studio 15 2017 Win64" -DUSE_CUDA=ON -DCMAKE_VERBOSE_MAKEFILE=ON -DGOOGLE_TEST=ON -DUSE_DMLC_GTEST=ON
+    """
+    bat """
+    cd build
+    "C:\\Program Files (x86)\\Microsoft Visual Studio\\2017\\Community\\MSBuild\\15.0\\Bin\\MSBuild.exe" xgboost.sln /m /p:Configuration=Release /nodeReuse:false
+    """
+    bat """
+    cd python-package
+    conda activate && python setup.py bdist_wheel --universal
+    """
+    echo "Insert vcomp140.dll (OpenMP runtime) into the wheel..."
+    bat """
+    cd python-package\\dist
+    COPY /B ..\\..\\tests\\ci_build\\insert_vcomp140.py
+    conda activate && python insert_vcomp140.py *.whl
+    """
+    echo 'Stashing Python wheel...'
+    stash name: 'xgboost_whl', includes: 'python-package/dist/*.whl'
+    archiveArtifacts artifacts: "python-package/dist/*.whl", allowEmptyArchive: true
+    echo 'Stashing C++ test executable (testxgboost)...'
+    stash name: 'xgboost_cpp_tests', includes: 'build/testxgboost.exe'
+    deleteDir()
+  }
+}
+
+def TestWin64CPU() {
+  node('win64 && cpu') {
+    unstash name: 'srcs'
+    unstash name: 'xgboost_whl'
+    echo "Test Win64 CPU"
+    echo "Installing Python wheel..."
+    bat "conda activate && (python -m pip uninstall -y xgboost || cd .)"
+    bat """
+    conda activate && for /R %%i in (python-package\\dist\\*.whl) DO python -m pip install "%%i"
+    """
+    echo "Running Python tests..."
+    bat "conda activate && python -m pytest -v -s --fulltrace tests\\python"
+    bat "conda activate && python -m pip uninstall -y xgboost"
+    deleteDir()
+  }
+}
+
+def TestWin64GPU(args) {
+  node("win64 && gpu && ${args.cuda_target}") {
+    unstash name: 'srcs'
+    unstash name: 'xgboost_whl'
+    unstash name: 'xgboost_cpp_tests'
+    echo "Test Win64 GPU (${args.cuda_target})"
+    bat "nvcc --version"
+    echo "Running C++ tests..."
+    bat "build\\testxgboost.exe"
+    echo "Installing Python wheel..."
+    bat "conda activate && (python -m pip uninstall -y xgboost || cd .)"
+    bat """
+    conda activate && for /R %%i in (python-package\\dist\\*.whl) DO python -m pip install "%%i"
+    """
+    echo "Running Python tests..."
+    bat """
+    conda activate && python -m pytest -v -s --fulltrace -m "(not slow) and (not mgpu)" tests\\python-gpu
+    """
+    bat "conda activate && python -m pip uninstall -y xgboost"
+    deleteDir()
+  }
+}

LICENSE

@@ -1,13 +1,201 @@
-Copyright (c) 2016 by Contributors
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
 
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
 
-   http://www.apache.org/licenses/LICENSE-2.0
+   1. Definitions.
 
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "{}"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright (c) 2018 by Contributors
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

Makefile

@@ -173,10 +173,14 @@ xgboost: $(CLI_OBJ) $(ALL_DEP)
 	$(CXX) $(CFLAGS) -o $@  $(filter %.o %.a, $^)  $(LDFLAGS)
 
 rcpplint:
-	python2 dmlc-core/scripts/lint.py xgboost ${LINT_LANG} R-package/src
+	python3 dmlc-core/scripts/lint.py xgboost ${LINT_LANG} R-package/src
 
 lint: rcpplint
-	python2 dmlc-core/scripts/lint.py xgboost ${LINT_LANG} include src plugin python-package
+	python3 dmlc-core/scripts/lint.py --exclude_path python-package/xgboost/dmlc-core \
+	  python-package/xgboost/include python-package/xgboost/lib \
+	  python-package/xgboost/make python-package/xgboost/rabit \
+	  python-package/xgboost/src --pylint-rc ${PWD}/python-package/.pylintrc xgboost \
+	  ${LINT_LANG} include src plugin python-package
 
 pylint:
 	flake8 --ignore E501 python-package
@@ -260,7 +264,8 @@ Rpack: clean_all
 	cp ./LICENSE xgboost
 	cat R-package/src/Makevars.in|sed '2s/.*/PKGROOT=./' | sed '3s/.*/ENABLE_STD_THREAD=0/' > xgboost/src/Makevars.in
 	cp xgboost/src/Makevars.in xgboost/src/Makevars.win
-	sed -i -e 's/@OPENMP_CXXFLAGS@/$$\(SHLIB_OPENMP_CFLAGS\)/g' xgboost/src/Makevars.win
+	sed -i -e 's/@OPENMP_CXXFLAGS@/$$\(SHLIB_OPENMP_CXXFLAGS\)/g' xgboost/src/Makevars.win
+	sed -i -e 's/-pthread/$$\(SHLIB_PTHREAD_FLAGS\)/g' xgboost/src/Makevars.win
 	bash R-package/remove_warning_suppression_pragma.sh
 	rm xgboost/remove_warning_suppression_pragma.sh
 

NEWS.md

@@ -3,6 +3,467 @@ XGBoost Change Log
 
 This file records the changes in xgboost library in reverse chronological order.
 
+## v0.90 (2019.05.18)
+
+### XGBoost Python package drops Python 2.x (#4379, #4381)
+Python 2.x is reaching its end-of-life at the end of this year. [Many scientific Python packages are now moving to drop Python 2.x](https://python3statement.org/).
+
+### XGBoost4J-Spark now requires Spark 2.4.x (#4377)
+* Spark 2.3 is reaching its end-of-life soon. See discussion at #4389.
+* **Consistent handling of missing values** (#4309, #4349, #4411): Many users had reported issue with inconsistent predictions between XGBoost4J-Spark and the Python XGBoost package. The issue was caused by Spark mis-handling non-zero missing values (NaN, -1, 999 etc). We now alert the user whenever Spark doesn't handle missing values correctly (#4309, #4349). See [the tutorial for dealing with missing values in XGBoost4J-Spark](https://xgboost.readthedocs.io/en/release_0.90/jvm/xgboost4j_spark_tutorial.html#dealing-with-missing-values). This fix also depends on the availability of Spark 2.4.x.
+
+### Roadmap: better performance scaling for multi-core CPUs (#4310)
+* Poor performance scaling of the `hist` algorithm for multi-core CPUs has been under investigation (#3810). #4310 optimizes quantile sketches and other pre-processing tasks. Special thanks to @SmirnovEgorRu.
+
+### Roadmap: Harden distributed training (#4250)
+* Make distributed training in XGBoost more robust by hardening [Rabit](https://github.com/dmlc/rabit), which implements [the AllReduce primitive](https://en.wikipedia.org/wiki/Reduce_%28parallel_pattern%29). In particular, improve test coverage on mechanisms for fault tolerance and recovery. Special thanks to @chenqin.
+
+### New feature: Multi-class metric functions for GPUs (#4368)
+* Metrics for multi-class classification have been ported to GPU: `merror`, `mlogloss`. Special thanks to @trivialfis.
+* With supported metrics, XGBoost will select the correct devices based on your system and `n_gpus` parameter.
+
+### New feature: Scikit-learn-like random forest API (#4148, #4255, #4258)
+* XGBoost Python package now offers `XGBRFClassifier` and `XGBRFRegressor` API to train random forests. See [the tutorial](https://xgboost.readthedocs.io/en/release_0.90/tutorials/rf.html). Special thanks to @canonizer
+
+### New feature: use external memory in GPU predictor (#4284, #4396, #4438, #4457)
+* It is now possible to make predictions on GPU when the input is read from external memory. This is useful when you want to make predictions with big dataset that does not fit into the GPU memory. Special thanks to @rongou, @canonizer, @sriramch.
+
+  ```python
+  dtest = xgboost.DMatrix('test_data.libsvm#dtest.cache')
+  bst.set_param('predictor', 'gpu_predictor')
+  bst.predict(dtest)
+  ```
+
+* Coming soon: GPU training (`gpu_hist`) with external memory
+
+### New feature: XGBoost can now handle comments in LIBSVM files (#4430)
+* Special thanks to @trivialfis and @hcho3
+
+### New feature: Embed XGBoost in your C/C++ applications using CMake (#4323, #4333, #4453)
+* It is now easier than ever to embed XGBoost in your C/C++ applications. In your CMakeLists.txt, add `xgboost::xgboost` as a linked library:
+
+  ```cmake
+  find_package(xgboost REQUIRED)
+  add_executable(api-demo c-api-demo.c)
+  target_link_libraries(api-demo xgboost::xgboost)
+  ```
+
+  [XGBoost C API documentation is available.](https://xgboost.readthedocs.io/en/release_0.90/dev) Special thanks to @trivialfis
+
+### Performance improvements
+* Use feature interaction constraints to narrow split search space (#4341, #4428)
+* Additional optimizations for `gpu_hist` (#4248, #4283)
+* Reduce OpenMP thread launches in `gpu_hist` (#4343)
+* Additional optimizations for multi-node multi-GPU random forests. (#4238)
+* Allocate unique prediction buffer for each input matrix, to avoid re-sizing GPU array (#4275)
+* Remove various synchronisations from CUDA API calls (#4205)
+* XGBoost4J-Spark
+  - Allow the user to control whether to cache partitioned training data, to potentially reduce execution time (#4268)
+
+### Bug-fixes
+* Fix node reuse in `hist` (#4404)
+* Fix GPU histogram allocation (#4347)
+* Fix matrix attributes not sliced (#4311)
+* Revise AUC and AUCPR metrics now work with weighted ranking task (#4216, #4436)
+* Fix timer invocation for InitDataOnce() in `gpu_hist` (#4206)
+* Fix R-devel errors (#4251)
+* Make gradient update in GPU linear updater thread-safe (#4259)
+* Prevent out-of-range access in column matrix (#4231)
+* Don't store DMatrix handle in Python object until it's initialized, to improve exception safety (#4317)
+* XGBoost4J-Spark
+  - Fix non-deterministic order within a zipped partition on prediction (#4388)
+  - Remove race condition on tracker shutdown (#4224)
+  - Allow set the parameter `maxLeaves`. (#4226)
+  - Allow partial evaluation of dataframe before prediction (#4407)
+  - Automatically set `maximize_evaluation_metrics` if not explicitly given (#4446)
+
+### API changes
+* Deprecate `reg:linear` in favor of `reg:squarederror`. (#4267, #4427)
+* Add attribute getter and setter to the Booster object in XGBoost4J (#4336)
+
+### Maintenance: Refactor C++ code for legibility and maintainability
+* Fix clang-tidy warnings. (#4149)
+* Remove deprecated C APIs. (#4266)
+* Use Monitor class to time functions in `hist`. (#4273)
+* Retire DVec class in favour of c++20 style span for device memory. (#4293)
+* Improve HostDeviceVector exception safety (#4301)
+
+### Maintenance: testing, continuous integration, build system
+* **Major refactor of CMakeLists.txt** (#4323, #4333, #4453): adopt modern CMake and export XGBoost as a target
+* **Major improvement in Jenkins CI pipeline** (#4234)
+  - Migrate all Linux tests to Jenkins (#4401)
+  - Builds and tests are now de-coupled, to test an artifact against multiple versions of CUDA, JDK, and other dependencies (#4401)
+  - Add Windows GPU to Jenkins CI pipeline (#4463, #4469)
+* Support CUDA 10.1 (#4223, #4232, #4265, #4468)
+* Python wheels are now built with CUDA 9.0, so that JIT is not required on Volta architecture (#4459)
+* Integrate with NVTX CUDA profiler (#4205)
+* Add a test for cpu predictor using external memory (#4308)
+* Refactor tests to get rid of duplication (#4358)
+* Remove test dependency on `craigcitro/r-travis`, since it's deprecated (#4353)
+* Add files from local R build to `.gitignore` (#4346)
+* Make XGBoost4J compatible with Java 9+ by revising NativeLibLoader (#4351)
+* Jenkins build for CUDA 10.0 (#4281)
+* Remove remaining `silent` and `debug_verbose` in Python tests (#4299)
+* Use all cores to build XGBoost4J lib on linux (#4304)
+* Upgrade Jenkins Linux build environment to GCC 5.3.1, CMake 3.6.0 (#4306)
+* Make CMakeLists.txt compatible with CMake 3.3 (#4420)
+* Add OpenMP option in CMakeLists.txt (#4339)
+* Get rid of a few trivial compiler warnings (#4312)
+* Add external Docker build cache, to speed up builds on Jenkins CI (#4331, #4334, #4458)
+* Fix Windows tests (#4403)
+* Fix a broken python test (#4395)
+* Use a fixed seed to split data in XGBoost4J-Spark tests, for reproducibility (#4417)
+* Add additional Python tests to test training under constraints (#4426)
+* Enable building with shared NCCL. (#4447)
+
+### Usability Improvements, Documentation
+* Document limitation of one-split-at-a-time Greedy tree learning heuristic (#4233)
+* Update build doc: PyPI wheel now support multi-GPU (#4219)
+* Fix docs for `num_parallel_tree` (#4221)
+* Fix document about `colsample_by*` parameter (#4340)
+* Make the train and test input with same colnames. (#4329)
+* Update R contribute link. (#4236)
+* Fix travis R tests (#4277)
+* Log version number in crash log in XGBoost4J-Spark (#4271, #4303)
+* Allow supression of Rabit output in Booster::train in XGBoost4J (#4262)
+* Add tutorial on handling missing values in XGBoost4J-Spark (#4425)
+* Fix typos (#4345, #4393, #4432, #4435)
+* Added language classifier in setup.py (#4327)
+* Added Travis CI badge (#4344)
+* Add BentoML to use case section (#4400)
+* Remove subtly sexist remark (#4418)
+* Add R vignette about parsing JSON dumps (#4439)
+
+### Acknowledgement
+**Contributors**: Nan Zhu (@CodingCat), Adam Pocock (@Craigacp), Daniel Hen (@Daniel8hen), Jiaxiang Li (@JiaxiangBU), Rory Mitchell (@RAMitchell), Egor Smirnov (@SmirnovEgorRu), Andy Adinets (@canonizer), Jonas (@elcombato), Harry Braviner (@harrybraviner), Philip Hyunsu Cho (@hcho3), Tong He (@hetong007), James Lamb (@jameslamb), Jean-Francois Zinque (@jeffzi), Yang Yang (@jokerkeny), Mayank Suman (@mayanksuman), jess (@monkeywithacupcake), Hajime Morrita (@omo), Ravi Kalia (@project-delphi), @ras44, Rong Ou (@rongou), Shaochen Shi (@shishaochen), Xu Xiao (@sperlingxx), @sriramch, Jiaming Yuan (@trivialfis), Christopher Suchanek (@wsuchy), Bozhao (@yubozhao)
+
+**Reviewers**: Nan Zhu (@CodingCat), Adam Pocock (@Craigacp), Daniel Hen (@Daniel8hen), Jiaxiang Li (@JiaxiangBU), Laurae (@Laurae2), Rory Mitchell (@RAMitchell), Egor Smirnov (@SmirnovEgorRu), @alois-bissuel, Andy Adinets (@canonizer), Chen Qin (@chenqin), Harry Braviner (@harrybraviner), Philip Hyunsu Cho (@hcho3), Tong He (@hetong007), @jakirkham, James Lamb (@jameslamb), Julien Schueller (@jschueller), Mayank Suman (@mayanksuman), Hajime Morrita (@omo), Rong Ou (@rongou), Sara Robinson (@sararob), Shaochen Shi (@shishaochen), Xu Xiao (@sperlingxx), @sriramch, Sean Owen (@srowen), Sergei Lebedev (@superbobry), Yuan (Terry) Tang (@terrytangyuan), Theodore Vasiloudis (@thvasilo), Matthew Tovbin (@tovbinm), Jiaming Yuan (@trivialfis), Xin Yin (@xydrolase)
+
+## v0.82 (2019.03.03)
+This release is packed with many new features and bug fixes.
+
+### Roadmap: better performance scaling for multi-core CPUs (#3957)
+* Poor performance scaling of the `hist` algorithm for multi-core CPUs has been under investigation (#3810). #3957 marks an important step toward better performance scaling, by using software pre-fetching and replacing STL vectors with C-style arrays. Special thanks to @Laurae2 and @SmirnovEgorRu.
+* See #3810 for latest progress on this roadmap.
+
+### New feature: Distributed Fast Histogram Algorithm (`hist`) (#4011, #4102, #4140, #4128)
+* It is now possible to run the `hist` algorithm in distributed setting. Special thanks to @CodingCat. The benefits include:
+  1. Faster local computation via feature binning
+  2. Support for monotonic constraints and feature interaction constraints
+  3. Simpler codebase than `approx`, allowing for future improvement
+* Depth-wise tree growing is now performed in a separate code path, so that cross-node syncronization is performed only once per level.
+
+### New feature: Multi-Node, Multi-GPU training (#4095)
+* Distributed training is now able to utilize clusters equipped with NVIDIA GPUs. In particular, the rabit AllReduce layer will communicate GPU device information. Special thanks to @mt-jones, @RAMitchell, @rongou, @trivialfis, @canonizer, and @jeffdk.
+* Resource management systems will be able to assign a rank for each GPU in the cluster.
+* In Dask, users will be able to construct a collection of XGBoost processes over an inhomogeneous device cluster (i.e. workers with different number and/or kinds of GPUs).
+
+### New feature: Multiple validation datasets in XGBoost4J-Spark (#3904, #3910)
+* You can now track the performance of the model during training with multiple evaluation datasets. By specifying `eval_sets` or call `setEvalSets` over a `XGBoostClassifier` or `XGBoostRegressor`, you can pass in multiple evaluation datasets typed as a `Map` from `String` to `DataFrame`. Special thanks to @CodingCat.
+* See the usage of multiple validation datasets [here](https://github.com/dmlc/xgboost/blob/0c1d5f1120c0a159f2567b267f0ec4ffadee00d0/jvm-packages/xgboost4j-example/src/main/scala/ml/dmlc/xgboost4j/scala/example/spark/SparkTraining.scala#L66-L78)
+
+### New feature: Additional metric functions for GPUs (#3952)
+* Element-wise metrics have been ported to GPU: `rmse`, `mae`, `logloss`, `poisson-nloglik`, `gamma-deviance`, `gamma-nloglik`, `error`, `tweedie-nloglik`. Special thanks to @trivialfis and @RAMitchell.
+* With supported metrics, XGBoost will select the correct devices based on your system and `n_gpus` parameter.
+
+### New feature: Column sampling at individual nodes (splits) (#3971)
+* Columns (features) can now be sampled at individual tree nodes, in addition to per-tree and per-level sampling. To enable per-node sampling, set `colsample_bynode` parameter, which represents the fraction of columns sampled at each node. This parameter is set to 1.0 by default (i.e. no sampling per node). Special thanks to @canonizer.
+* The `colsample_bynode` parameter works cumulatively with other `colsample_by*` parameters: for example, `{'colsample_bynode':0.5, 'colsample_bytree':0.5}` with 100 columns will give 25 features to choose from at each split.
+
+### Major API change: consistent logging level via `verbosity` (#3982, #4002, #4138)
+* XGBoost now allows fine-grained control over logging. You can set `verbosity` to 0 (silent), 1 (warning), 2 (info), and 3 (debug). This is useful for controlling the amount of logging outputs. Special thanks to @trivialfis.
+* Parameters `silent` and `debug_verbose` are now deprecated.
+* Note: Sometimes XGBoost tries to change configurations based on heuristics, which is displayed as warning message.  If there's unexpected behaviour, please try to increase value of verbosity.
+
+### Major bug fix: external memory (#4040, #4193)
+* Clarify object ownership in multi-threaded prefetcher, to avoid memory error.
+* Correctly merge two column batches (which uses [CSC layout](https://en.wikipedia.org/wiki/Sparse_matrix#Compressed_sparse_column_(CSC_or_CCS))).
+* Add unit tests for external memory.
+* Special thanks to @trivialfis and @hcho3.
+
+### Major bug fix: early stopping fixed in XGBoost4J and XGBoost4J-Spark (#3928, #4176)
+* Early stopping in XGBoost4J and XGBoost4J-Spark is now consistent with its counterpart in the Python package. Training stops if the current iteration is `earlyStoppingSteps` away from the best iteration. If there are multiple evaluation sets, only the last one is used to determinate early stop.
+* See the updated documentation [here](https://xgboost.readthedocs.io/en/release_0.82/jvm/xgboost4j_spark_tutorial.html#early-stopping)
+* Special thanks to @CodingCat, @yanboliang, and @mingyang.
+
+### Major bug fix: infrequent features should not crash distributed training (#4045)
+* For infrequently occuring features, some partitions may not get any instance. This scenario used to crash distributed training due to mal-formed ranges. The problem has now been fixed.
+* In practice, one-hot-encoded categorical variables tend to produce rare features, particularly when the cardinality is high.
+* Special thanks to @CodingCat.
+
+### Performance improvements
+* Faster, more space-efficient radix sorting in `gpu_hist` (#3895)
+* Subtraction trick in histogram calculation in `gpu_hist` (#3945)
+* More performant re-partition in XGBoost4J-Spark (#4049)
+
+### Bug-fixes
+* Fix semantics of `gpu_id` when running multiple XGBoost processes on a multi-GPU machine (#3851)
+* Fix page storage path for external memory on Windows (#3869)
+* Fix configuration setup so that DART utilizes GPU (#4024)
+* Eliminate NAN values from SHAP prediction (#3943)
+* Prevent empty quantile sketches in `hist` (#4155)
+* Enable running objectives with 0 GPU (#3878)
+* Parameters are no longer dependent on system locale (#3891, #3907)
+* Use consistent data type in the GPU coordinate descent code (#3917)
+* Remove undefined behavior in the CLI config parser on the ARM platform (#3976)
+* Initialize counters in GPU AllReduce (#3987)
+* Prevent deadlocks in GPU AllReduce (#4113)
+* Load correct values from sliced NumPy arrays (#4147, #4165)
+* Fix incorrect GPU device selection (#4161)
+* Make feature binning logic in `hist` aware of query groups when running a ranking task (#4115). For ranking task, query groups are weighted, not individual instances.
+* Generate correct C++ exception type for `LOG(FATAL)` macro (#4159)
+* Python package
+  - Python package should run on system without `PATH` environment variable (#3845)
+  - Fix `coef_` and `intercept_` signature to be compatible with `sklearn.RFECV` (#3873)
+  - Use UTF-8 encoding in Python package README, to support non-English locale (#3867)
+  - Add AUC-PR to list of metrics to maximize for early stopping (#3936)
+  - Allow loading pickles without `self.booster` attribute, for backward compatibility (#3938, #3944)
+  - White-list DART for feature importances (#4073)
+  - Update usage of [h2oai/datatable](https://github.com/h2oai/datatable) (#4123)
+* XGBoost4J-Spark
+  - Address scalability issue in prediction (#4033)
+  - Enforce the use of per-group weights for ranking task (#4118)
+  - Fix vector size of `rawPredictionCol` in `XGBoostClassificationModel` (#3932)
+  - More robust error handling in Spark tracker (#4046, #4108)
+  - Fix return type of `setEvalSets` (#4105)
+  - Return correct value of `getMaxLeaves` (#4114)
+
+### API changes
+* Add experimental parameter `single_precision_histogram` to use single-precision histograms for the `gpu_hist` algorithm (#3965)
+* Python package
+  - Add option to select type of feature importances in the scikit-learn inferface (#3876)
+  - Add `trees_to_df()` method to dump decision trees as Pandas data frame (#4153)
+  - Add options to control node shapes in the GraphViz plotting function (#3859)
+  - Add `xgb_model` option to `XGBClassifier`, to load previously saved model (#4092)
+  - Passing lists into `DMatrix` is now deprecated (#3970)
+* XGBoost4J
+  - Support multiple feature importance features (#3801)
+
+### Maintenance: Refactor C++ code for legibility and maintainability
+* Refactor `hist` algorithm code and add unit tests (#3836)
+* Minor refactoring of split evaluator in `gpu_hist` (#3889)
+* Removed unused leaf vector field in the tree model (#3989)
+* Simplify the tree representation by combining `TreeModel` and `RegTree` classes (#3995)
+* Simplify and harden tree expansion code (#4008, #4015)
+* De-duplicate parameter classes in the linear model algorithms (#4013)
+* Robust handling of ranges with C++20 span in `gpu_exact` and `gpu_coord_descent` (#4020, #4029)
+* Simplify tree training code (#3825). Also use Span class for robust handling of ranges.
+
+### Maintenance: testing, continuous integration, build system
+* Disallow `std::regex` since it's not supported by GCC 4.8.x (#3870)
+* Add multi-GPU tests for coordinate descent algorithm for linear models (#3893, #3974)
+* Enforce naming style in Python lint (#3896)
+* Refactor Python tests (#3897, #3901): Use pytest exclusively, display full trace upon failure
+* Address `DeprecationWarning` when using Python collections (#3909)
+* Use correct group for maven site plugin (#3937)
+* Jenkins CI is now using on-demand EC2 instances exclusively, due to unreliability of Spot instances (#3948)
+* Better GPU performance logging (#3945)
+* Fix GPU tests on machines with only 1 GPU (#4053)
+* Eliminate CRAN check warnings and notes (#3988)
+* Add unit tests for tree serialization (#3989)
+* Add unit tests for tree fitting functions in `hist` (#4155)
+* Add a unit test for `gpu_exact` algorithm (#4020)
+* Correct JVM CMake GPU flag (#4071)
+* Fix failing Travis CI on Mac (#4086)
+* Speed up Jenkins by not compiling CMake (#4099)
+* Analyze C++ and CUDA code using clang-tidy, as part of Jenkins CI pipeline (#4034)
+* Fix broken R test: Install Homebrew GCC (#4142)
+* Check for empty datasets in GPU unit tests (#4151)
+* Fix Windows compilation (#4139)
+* Comply with latest convention of cpplint (#4157)
+* Fix a unit test in `gpu_hist` (#4158)
+* Speed up data generation in Python tests (#4164)
+
+### Usability Improvements
+* Add link to [InfoWorld 2019 Technology of the Year Award](https://www.infoworld.com/article/3336072/application-development/infoworlds-2019-technology-of-the-year-award-winners.html) (#4116)
+* Remove outdated AWS YARN tutorial (#3885)
+* Document current limitation in number of features (#3886)
+* Remove unnecessary warning when `gblinear` is selected (#3888)
+* Document limitation of CSV parser: header not supported (#3934)
+* Log training parameters in XGBoost4J-Spark (#4091)
+* Clarify early stopping behavior in the scikit-learn interface (#3967)
+* Clarify behavior of `max_depth` parameter (#4078)
+* Revise Python docstrings for ranking task (#4121). In particular, weights must be per-group in learning-to-rank setting.
+* Document parameter `num_parallel_tree` (#4022)
+* Add Jenkins status badge (#4090)
+* Warn users against using internal functions of `Booster` object (#4066)
+* Reformat `benchmark_tree.py` to comply with Python style convention (#4126)
+* Clarify a comment in `objectiveTrait` (#4174)
+* Fix typos and broken links in documentation (#3890, #3872, #3902, #3919, #3975, #4027, #4156, #4167)
+
+### Acknowledgement
+**Contributors** (in no particular order): Jiaming Yuan (@trivialfis), Hyunsu Cho (@hcho3), Nan Zhu (@CodingCat), Rory Mitchell (@RAMitchell), Yanbo Liang (@yanboliang), Andy Adinets (@canonizer), Tong He (@hetong007), Yuan Tang (@terrytangyuan)
+
+**First-time Contributors** (in no particular order): Jelle Zijlstra (@JelleZijlstra), Jiacheng Xu (@jiachengxu), @ajing, Kashif Rasul (@kashif), @theycallhimavi, Joey Gao (@pjgao), Prabakaran Kumaresshan (@nixphix), Huafeng Wang (@huafengw), @lyxthe, Sam Wilkinson (@scwilkinson), Tatsuhito Kato (@stabacov), Shayak Banerjee (@shayakbanerjee), Kodi Arfer (@Kodiologist), @KyleLi1985, Egor Smirnov (@SmirnovEgorRu), @tmitanitky, Pasha Stetsenko (@st-pasha), Kenichi Nagahara (@keni-chi), Abhai Kollara Dilip (@abhaikollara), Patrick Ford (@pford221), @hshujuan, Matthew Jones (@mt-jones), Thejaswi Rao (@teju85), Adam November (@anovember)
+
+**First-time Reviewers** (in no particular order): Mingyang Hu (@mingyang), Theodore Vasiloudis (@thvasilo), Jakub Troszok (@troszok), Rong Ou (@rongou), @Denisevi4, Matthew Jones (@mt-jones), Jeff Kaplan (@jeffdk)
+
+## v0.81 (2018.11.04)
+### New feature: feature interaction constraints
+* Users are now able to control which features (independent variables) are allowed to interact by specifying feature interaction constraints (#3466).
+* [Tutorial](https://xgboost.readthedocs.io/en/release_0.81/tutorials/feature_interaction_constraint.html) is available, as well as [R](https://github.com/dmlc/xgboost/blob/9254c58e4dfff6a59dc0829a2ceb02e45ed17cd0/R-package/demo/interaction_constraints.R) and [Python](https://github.com/dmlc/xgboost/blob/9254c58e4dfff6a59dc0829a2ceb02e45ed17cd0/tests/python/test_interaction_constraints.py) examples.
+
+### New feature: learning to rank using scikit-learn interface
+* Learning to rank task is now available for the scikit-learn interface of the Python package (#3560, #3848). It is now possible to integrate the XGBoost ranking model into the scikit-learn learning pipeline.
+* Examples of using `XGBRanker` class is found at [demo/rank/rank_sklearn.py](https://github.com/dmlc/xgboost/blob/24a268a2e3cb17302db3d72da8f04016b7d352d9/demo/rank/rank_sklearn.py).
+
+### New feature: R interface for SHAP interactions
+* SHAP (SHapley Additive exPlanations) is a unified approach to explain the output of any machine learning model. Previously, this feature was only available from the Python package; now it is available from the R package as well (#3636).
+
+### New feature: GPU predictor now use multiple GPUs to predict
+* GPU predictor is now able to utilize multiple GPUs at once to accelerate prediction (#3738)
+
+### New feature: Scale distributed XGBoost to large-scale clusters
+* Fix OS file descriptor limit assertion error on large cluster (#3835, dmlc/rabit#73) by replacing `select()` based AllReduce/Broadcast with `poll()` based implementation.
+* Mitigate tracker "thundering herd" issue on large cluster. Add exponential backoff retry when workers connect to tracker.
+* With this change, we were able to scale to 1.5k executors on a 12 billion row dataset after some tweaks here and there.
+
+### New feature: Additional objective functions for GPUs
+* New objective functions ported to GPU: `hinge`, `multi:softmax`, `multi:softprob`, `count:poisson`, `reg:gamma`, `"reg:tweedie`.
+* With supported objectives, XGBoost will select the correct devices based on your system and `n_gpus` parameter.
+
+### Major bug fix: learning to rank with XGBoost4J-Spark
+* Previously, `repartitionForData` would shuffle data and lose ordering necessary for ranking task.
+* To fix this issue, data points within each RDD partition is explicitly group by their group (query session) IDs (#3654). Also handle empty RDD partition carefully (#3750).
+
+### Major bug fix: early stopping fixed in XGBoost4J-Spark
+* Earlier implementation of early stopping had incorrect semantics and didn't let users to specify direction for optimizing (maximize / minimize)
+* A parameter `maximize_evaluation_metrics` is defined so as to tell whether a metric should be maximized or minimized as part of early stopping criteria (#3808). Also early stopping now has correct semantics.
+
+### API changes
+* Column sampling by level (`colsample_bylevel`) is now functional for `hist` algorithm (#3635, #3862)
+* GPU tag `gpu:` for regression objectives are now deprecated. XGBoost will select the correct devices automatically (#3643)
+* Add `disable_default_eval_metric` parameter to disable default metric (#3606)
+* Experimental AVX support for gradient computation is removed (#3752)
+* XGBoost4J-Spark
+  - Add `rank:ndcg` and `rank:map` to supported objectives (#3697)
+* Python package
+  - Add `callbacks` argument to `fit()` function of sciki-learn API (#3682)
+  - Add `XGBRanker` to scikit-learn interface (#3560, #3848)
+  - Add `validate_features` argument to `predict()` function of scikit-learn API (#3653)
+  - Allow scikit-learn grid search over parameters specified as keyword arguments (#3791)
+  - Add `coef_` and `intercept_` as properties of scikit-learn wrapper (#3855). Some scikit-learn functions expect these properties.
+
+### Performance improvements
+* Address very high GPU memory usage for large data (#3635)
+* Fix performance regression within `EvaluateSplits()` of `gpu_hist` algorithm. (#3680)
+
+### Bug-fixes
+* Fix a problem in GPU quantile sketch with tiny instance weights. (#3628)
+* Fix copy constructor for `HostDeviceVectorImpl` to prevent dangling pointers (#3657)
+* Fix a bug in partitioned file loading (#3673)
+* Fixed an uninitialized pointer in `gpu_hist` (#3703)
+* Reshared data among GPUs when number of GPUs is changed (#3721)
+* Add back `max_delta_step` to split evaluation (#3668)
+* Do not round up integer thresholds for integer features in JSON dump (#3717)
+* Use `dmlc::TemporaryDirectory` to handle temporaries in cross-platform way (#3783)
+* Fix accuracy problem with `gpu_hist` when `min_child_weight` and `lambda` are set to 0 (#3793)
+* Make sure that `tree_method` parameter is recognized and not silently ignored (#3849)
+* XGBoost4J-Spark
+  - Make sure `thresholds` are considered when executing `predict()` method (#3577)
+  - Avoid losing precision when computing probabilities by converting to `Double` early (#3576)
+  - `getTreeLimit()` should return `Int` (#3602)
+  - Fix checkpoint serialization on HDFS (#3614)
+  - Throw `ControlThrowable` instead of `InterruptedException` so that it is properly re-thrown (#3632)
+  - Remove extraneous output to stdout (#3665)
+  - Allow specification of task type for custom objectives and evaluations (#3646)
+  - Fix distributed updater check (#3739)
+  - Fix issue when spark job execution thread cannot return before we execute `first()` (#3758)
+* Python package
+  - Fix accessing `DMatrix.handle` before it is set (#3599)
+  - `XGBClassifier.predict()` should return margin scores when `output_margin` is set to true (#3651)
+  - Early stopping callback should maximize metric of form `NDCG@n-` (#3685)
+  - Preserve feature names when slicing `DMatrix` (#3766)
+* R package
+  - Replace `nround` with `nrounds` to match actual parameter (#3592)
+  - Amend `xgb.createFolds` to handle classes of a single element (#3630)
+  - Fix buggy random generator and make `colsample_bytree` functional (#3781)
+
+### Maintenance: testing, continuous integration, build system
+* Add sanitizers tests to Travis CI (#3557)
+* Add NumPy, Matplotlib, Graphviz as requirements for doc build (#3669)
+* Comply with CRAN submission policy (#3660, #3728)
+* Remove copy-paste error in JVM test suite (#3692)
+* Disable flaky tests in `R-package/tests/testthat/test_update.R` (#3723)
+* Make Python tests compatible with scikit-learn 0.20 release (#3731)
+* Separate out restricted and unrestricted tasks, so that pull requests don't build downloadable artifacts (#3736)
+* Add multi-GPU unit test environment (#3741)
+* Allow plug-ins to be built by CMake (#3752)
+* Test wheel compatibility on CPU containers for pull requests (#3762)
+* Fix broken doc build due to Matplotlib 3.0 release (#3764)
+* Produce `xgboost.so` for XGBoost-R on Mac OSX, so that `make install` works (#3767)
+* Retry Jenkins CI tests up to 3 times to improve reliability (#3769, #3769, #3775, #3776, #3777)
+* Add basic unit tests for `gpu_hist` algorithm (#3785)
+* Fix Python environment for distributed unit tests (#3806)
+* Test wheels on CUDA 10.0 container for compatibility (#3838)
+* Fix JVM doc build (#3853)
+
+### Maintenance: Refactor C++ code for legibility and maintainability
+* Merge generic device helper functions into `GPUSet` class (#3626)
+* Re-factor column sampling logic into `ColumnSampler` class (#3635, #3637)
+* Replace `std::vector` with `HostDeviceVector` in `MetaInfo` and `SparsePage` (#3446)
+* Simplify `DMatrix` class (#3395)
+* De-duplicate CPU/GPU code using `Transform` class (#3643, #3751)
+* Remove obsoleted `QuantileHistMaker` class (#3761)
+* Remove obsoleted `NoConstraint` class (#3792)
+
+### Other Features
+* C++20-compliant Span class for safe pointer indexing (#3548, #3588)
+* Add helper functions to manipulate multiple GPU devices (#3693)
+* XGBoost4J-Spark
+  - Allow specifying host ip from the `xgboost-tracker.properties file` (#3833). This comes in handy when `hosts` files doesn't correctly define localhost.
+
+### Usability Improvements
+* Add reference to GitHub repository in `pom.xml` of JVM packages (#3589)
+* Add R demo of multi-class classification (#3695)
+* Document JSON dump functionality (#3600, #3603)
+* Document CUDA requirement and lack of external memory for GPU algorithms (#3624)
+* Document LambdaMART objectives, both pairwise and listwise (#3672)
+* Document `aucpr` evaluation metric (#3687)
+* Document gblinear parameters: `feature_selector` and `top_k` (#3780)
+* Add instructions for using MinGW-built XGBoost with Python. (#3774)
+* Removed nonexistent parameter `use_buffer` from documentation (#3610)
+* Update Python API doc to include all classes and members (#3619, #3682)
+* Fix typos and broken links in documentation (#3618, #3640, #3676, #3713, #3759, #3784, #3843, #3852)
+* Binary classification demo should produce LIBSVM with 0-based indexing (#3652)
+* Process data once for Python and CLI examples of learning to rank (#3666)
+* Include full text of Apache 2.0 license in the repository (#3698)
+* Save predictor parameters in model file (#3856)
+* JVM packages
+  - Let users specify feature names when calling `getModelDump` and `getFeatureScore` (#3733)
+  - Warn the user about the lack of over-the-wire encryption (#3667)
+  - Fix errors in examples (#3719)
+  - Document choice of trackers (#3831)
+  - Document that vanilla Apache Spark is required (#3854)
+* Python package
+  - Document that custom objective can't contain colon (:) (#3601)
+  - Show a better error message for failed library loading (#3690)
+  - Document that feature importance is unavailable for non-tree learners (#3765)
+  - Document behavior of `get_fscore()` for zero-importance features (#3763)
+  - Recommend pickling as the way to save `XGBClassifier` / `XGBRegressor` / `XGBRanker` (#3829)
+* R package
+  - Enlarge variable importance plot to make it more visible (#3820)
+
+### BREAKING CHANGES
+* External memory page files have changed, breaking backwards compatibility for temporary storage used during external memory training. This only affects external memory users upgrading their xgboost version - we recommend clearing all `*.page` files before resuming training. Model serialization is unaffected.
+
+### Known issues
+* Quantile sketcher fails to produce any quantile for some edge cases (#2943)
+* The `hist` algorithm leaks memory when used with learning rate decay callback (#3579)
+* Using custom evaluation funciton together with early stopping causes assertion failure in XGBoost4J-Spark (#3595)
+* Early stopping doesn't work with `gblinear` learner (#3789)
+* Label and weight vectors are not reshared upon the change in number of GPUs (#3794). To get around this issue, delete the `DMatrix` object and re-load.
+* The `DMatrix` Python objects are initialized with incorrect values when given array slices (#3841)
+* The `gpu_id` parameter is broken and not yet properly supported (#3850)
+
+### Acknowledgement
+**Contributors** (in no particular order): Hyunsu Cho (@hcho3), Jiaming Yuan (@trivialfis), Nan Zhu (@CodingCat), Rory Mitchell (@RAMitchell), Andy Adinets (@canonizer), Vadim Khotilovich (@khotilov), Sergei Lebedev (@superbobry)
+
+**First-time Contributors** (in no particular order): Matthew Tovbin (@tovbinm), Jakob Richter (@jakob-r), Grace Lam (@grace-lam), Grant W Schneider (@grantschneider), Andrew Thia (@BlueTea88), Sergei Chipiga (@schipiga), Joseph Bradley (@jkbradley), Chen Qin (@chenqin), Jerry Lin (@linjer), Dmitriy Rybalko (@rdtft), Michael Mui (@mmui), Takahiro Kojima (@515hikaru), Bruce Zhao (@BruceZhaoR), Wei Tian (@weitian), Saumya Bhatnagar (@Sam1301), Juzer Shakir (@JuzerShakir), Zhao Hang (@cleghom), Jonathan Friedman (@jontonsoup), Bruno Tremblay (@meztez), Boris Filippov (@frenzykryger), @Shiki-H, @mrgutkun, @gorogm, @htgeis, @jakehoare, @zengxy, @KOLANICH
+
+**First-time Reviewers** (in no particular order): Nikita Titov (@StrikerRUS), Xiangrui Meng (@mengxr), Nirmal Borah (@Nirmal-Neel)
+
+
 ## v0.80 (2018.08.13)
 * **JVM packages received a major upgrade**: To consolidate the APIs and improve the user experience, we refactored the design of XGBoost4J-Spark in a significant manner. (#3387)
   - Consolidated APIs: It is now much easier to integrate XGBoost models into a Spark ML pipeline. Users can control behaviors like output leaf prediction results by setting corresponding column names. Training is now more consistent with other Estimators in Spark MLLIB: there is now one single method `fit()` to train decision trees.
@@ -13,7 +474,7 @@ This file records the changes in xgboost library in reverse chronological order.
   - Latest master: https://xgboost.readthedocs.io/en/latest
   - 0.80 stable: https://xgboost.readthedocs.io/en/release_0.80
   - 0.72 stable: https://xgboost.readthedocs.io/en/release_0.72
-* Ranking task now uses instance weights (#3379)
+* Support for per-group weights in ranking objective (#3379)
 * Fix inaccurate decimal parsing (#3546)
 * New functionality
   - Query ID column support in LIBSVM data files (#2749). This is convenient for performing ranking task in distributed setting.
@@ -173,7 +634,7 @@ This version is only applicable for the Python package. The content is identical
   - Compatibility fix for Python 2.6
   - Call `print_evaluation` callback at last iteration
   - Use appropriate integer types when calling native code, to prevent truncation and memory error
-  - Fix shared library loading on Mac OS X 
+  - Fix shared library loading on Mac OS X
 * R package:
   - New parameters:
     - `silent` in `xgb.DMatrix()`
@@ -214,7 +675,7 @@ This version is only applicable for the Python package. The content is identical
   - Support instance weights
   - Use `SparkParallelismTracker` to prevent jobs from hanging forever
   - Expose train-time evaluation metrics via `XGBoostModel.summary`
-  - Option to specify `host-ip` explicitly in the Rabit tracker 
+  - Option to specify `host-ip` explicitly in the Rabit tracker
 * Documentation
   - Better math notation for gradient boosting
   - Updated build instructions for Mac OS X

R-package/CMakeLists.txt

@@ -0,0 +1,34 @@
+find_package(LibR REQUIRED)
+message(STATUS "LIBR_CORE_LIBRARY " ${LIBR_CORE_LIBRARY})
+
+file(GLOB_RECURSE R_SOURCES
+  ${CMAKE_CURRENT_LIST_DIR}/src/*.cc
+  ${CMAKE_CURRENT_LIST_DIR}/src/*.c)
+# Use object library to expose symbols
+add_library(xgboost-r OBJECT ${R_SOURCES})
+
+set(R_DEFINITIONS
+  -DXGBOOST_STRICT_R_MODE=1
+  -DXGBOOST_CUSTOMIZE_GLOBAL_PRNG=1
+  -DDMLC_LOG_BEFORE_THROW=0
+  -DDMLC_DISABLE_STDIN=1
+  -DDMLC_LOG_CUSTOMIZE=1
+  -DRABIT_CUSTOMIZE_MSG_
+  -DRABIT_STRICT_CXX98_)
+target_compile_definitions(xgboost-r
+  PRIVATE ${R_DEFINITIONS})
+target_include_directories(xgboost-r
+  PRIVATE
+  ${LIBR_INCLUDE_DIRS}
+  ${PROJECT_SOURCE_DIR}/include
+  ${PROJECT_SOURCE_DIR}/dmlc-core/include
+  ${PROJECT_SOURCE_DIR}/rabit/include)
+set_target_properties(
+  xgboost-r PROPERTIES
+  CXX_STANDARD 11
+  CXX_STANDARD_REQUIRED ON
+  POSITION_INDEPENDENT_CODE ON)
+
+set(XGBOOST_DEFINITIONS ${R_DEFINITIONS} PARENT_SCOPE)
+set(XGBOOST_OBJ_SOURCES $<TARGET_OBJECTS:xgboost-r> PARENT_SCOPE)
+set(LINKED_LIBRARIES_PRIVATE ${LINKED_LIBRARIES_PRIVATE} ${LIBR_CORE_LIBRARY} PARENT_SCOPE)

R-package/DESCRIPTION

@@ -1,8 +1,8 @@
 Package: xgboost
 Type: Package
 Title: Extreme Gradient Boosting
-Version: 0.80.1
-Date: 2018-08-13
+Version: 0.90.0.1
+Date: 2019-05-18
 Authors@R: c(
   person("Tianqi", "Chen", role = c("aut"),
          email = "tianqi.tchen@gmail.com"),
@@ -52,7 +52,9 @@ Suggests:
     vcd (>= 1.3),
     testthat,
     lintr,
-    igraph (>= 1.0.1)
+    igraph (>= 1.0.1),
+    jsonlite,
+    float
 Depends:
     R (>= 3.3.0)
 Imports:
@@ -61,5 +63,5 @@ Imports:
     data.table (>= 1.9.6),
     magrittr (>= 1.5),
     stringi (>= 0.5.2)
-RoxygenNote: 6.0.1
+RoxygenNote: 6.1.0
 SystemRequirements: GNU make, C++11

R-package/R/callbacks.R

@@ -1,26 +1,26 @@
 #' Callback closures for booster training.
 #'
 #' These are used to perform various service tasks either during boosting iterations or at the end.
-#' This approach helps to modularize many of such tasks without bloating the main training methods, 
+#' This approach helps to modularize many of such tasks without bloating the main training methods,
 #' and it offers .
-#' 
+#'
 #' @details
 #' By default, a callback function is run after each boosting iteration.
 #' An R-attribute \code{is_pre_iteration} could be set for a callback to define a pre-iteration function.
-#' 
-#' When a callback function has \code{finalize} parameter, its finalizer part will also be run after 
+#'
+#' When a callback function has \code{finalize} parameter, its finalizer part will also be run after
 #' the boosting is completed.
-#' 
-#' WARNING: side-effects!!! Be aware that these callback functions access and modify things in 
+#'
+#' WARNING: side-effects!!! Be aware that these callback functions access and modify things in
 #' the environment from which they are called from, which is a fairly uncommon thing to do in R.
-#' 
-#' To write a custom callback closure, make sure you first understand the main concepts about R envoronments.
-#' Check either R documentation on \code{\link[base]{environment}} or the 
-#' \href{http://adv-r.had.co.nz/Environments.html}{Environments chapter} from the "Advanced R" 
+#'
+#' To write a custom callback closure, make sure you first understand the main concepts about R environments.
+#' Check either R documentation on \code{\link[base]{environment}} or the
+#' \href{http://adv-r.had.co.nz/Environments.html}{Environments chapter} from the "Advanced R"
 #' book by Hadley Wickham. Further, the best option is to read the code of some of the existing callbacks -
-#' choose ones that do something similar to what you want to achieve. Also, you would need to get familiar 
+#' choose ones that do something similar to what you want to achieve. Also, you would need to get familiar
 #' with the objects available inside of the \code{xgb.train} and \code{xgb.cv} internal environments.
-#' 
+#'
 #' @seealso
 #' \code{\link{cb.print.evaluation}},
 #' \code{\link{cb.evaluation.log}},
@@ -30,42 +30,42 @@
 #' \code{\link{cb.cv.predict}},
 #' \code{\link{xgb.train}},
 #' \code{\link{xgb.cv}}
-#' 
+#'
 #' @name callbacks
 NULL
 
 #
 # Callbacks -------------------------------------------------------------------
-# 
+#
 
 #' Callback closure for printing the result of evaluation
-#' 
+#'
 #' @param period  results would be printed every number of periods
 #' @param showsd  whether standard deviations should be printed (when available)
-#' 
+#'
 #' @details
 #' The callback function prints the result of evaluation at every \code{period} iterations.
 #' The initial and the last iteration's evaluations are always printed.
-#' 
+#'
 #' Callback function expects the following values to be set in its calling frame:
 #' \code{bst_evaluation} (also \code{bst_evaluation_err} when available),
 #' \code{iteration},
 #' \code{begin_iteration},
 #' \code{end_iteration}.
-#' 
+#'
 #' @seealso
 #' \code{\link{callbacks}}
-#' 
+#'
 #' @export
 cb.print.evaluation <- function(period = 1, showsd = TRUE) {
-  
+
   callback <- function(env = parent.frame()) {
     if (length(env$bst_evaluation) == 0 ||
         period == 0 ||
         NVL(env$rank, 0) != 0 )
       return()
-    
-    i <- env$iteration 
+
+    i <- env$iteration
     if ((i-1) %% period == 0 ||
         i == env$begin_iteration ||
         i == env$end_iteration) {
@@ -81,48 +81,48 @@ cb.print.evaluation <- function(period = 1, showsd = TRUE) {
 
 
 #' Callback closure for logging the evaluation history
-#' 
+#'
 #' @details
 #' This callback function appends the current iteration evaluation results \code{bst_evaluation}
 #' available in the calling parent frame to the \code{evaluation_log} list in a calling frame.
-#' 
-#' The finalizer callback (called with \code{finalize = TURE} in the end) converts 
+#'
+#' The finalizer callback (called with \code{finalize = TURE} in the end) converts
 #' the \code{evaluation_log} list into a final data.table.
-#' 
-#' The iteration evaluation result \code{bst_evaluation} must be a named numeric vector. 
-#' 
-#' Note: in the column names of the final data.table, the dash '-' character is replaced with 
+#'
+#' The iteration evaluation result \code{bst_evaluation} must be a named numeric vector.
+#'
+#' Note: in the column names of the final data.table, the dash '-' character is replaced with
 #' the underscore '_' in order to make the column names more like regular R identifiers.
-#' 
+#'
 #' Callback function expects the following values to be set in its calling frame:
 #' \code{evaluation_log},
 #' \code{bst_evaluation},
 #' \code{iteration}.
-#' 
+#'
 #' @seealso
 #' \code{\link{callbacks}}
-#' 
+#'
 #' @export
 cb.evaluation.log <- function() {
 
   mnames <- NULL
-  
+
   init <- function(env) {
     if (!is.list(env$evaluation_log))
       stop("'evaluation_log' has to be a list")
     mnames <<- names(env$bst_evaluation)
     if (is.null(mnames) || any(mnames == ""))
       stop("bst_evaluation must have non-empty names")
-    
+
     mnames <<- gsub('-', '_', names(env$bst_evaluation))
     if(!is.null(env$bst_evaluation_err))
       mnames <<- c(paste0(mnames, '_mean'), paste0(mnames, '_std'))
   }
-  
+
   finalizer <- function(env) {
     env$evaluation_log <- as.data.table(t(simplify2array(env$evaluation_log)))
     setnames(env$evaluation_log, c('iter', mnames))
-    
+
     if(!is.null(env$bst_evaluation_err)) {
       # rearrange col order from _mean,_mean,...,_std,_std,...
       # to be _mean,_std,_mean,_std,...
@@ -135,18 +135,18 @@ cb.evaluation.log <- function() {
       env$evaluation_log <- env$evaluation_log[, c('iter', cnames), with = FALSE]
     }
   }
-  
+
   callback <- function(env = parent.frame(), finalize = FALSE) {
     if (is.null(mnames))
       init(env)
 
     if (finalize)
       return(finalizer(env))
-    
+
     ev <- env$bst_evaluation
     if(!is.null(env$bst_evaluation_err))
       ev <- c(ev, env$bst_evaluation_err)
-    env$evaluation_log <- c(env$evaluation_log, 
+    env$evaluation_log <- c(env$evaluation_log,
                             list(c(iter = env$iteration, ev)))
   }
   attr(callback, 'call') <- match.call()
@@ -154,21 +154,21 @@ cb.evaluation.log <- function() {
   callback
 }
 
-#' Callback closure for restetting the booster's parameters at each iteration.
-#' 
+#' Callback closure for resetting the booster's parameters at each iteration.
+#'
 #' @param new_params a list where each element corresponds to a parameter that needs to be reset.
-#'        Each element's value must be either a vector of values of length \code{nrounds} 
-#'        to be set at each iteration, 
-#'        or a function of two parameters \code{learning_rates(iteration, nrounds)} 
-#'        which returns a new parameter value by using the current iteration number 
+#'        Each element's value must be either a vector of values of length \code{nrounds}
+#'        to be set at each iteration,
+#'        or a function of two parameters \code{learning_rates(iteration, nrounds)}
+#'        which returns a new parameter value by using the current iteration number
 #'        and the total number of boosting rounds.
-#' 
-#' @details 
+#'
+#' @details
 #' This is a "pre-iteration" callback function used to reset booster's parameters
 #' at the beginning of each iteration.
-#' 
-#' Note that when training is resumed from some previous model, and a function is used to 
-#' reset a parameter value, the \code{nround} argument in this function would be the 
+#'
+#' Note that when training is resumed from some previous model, and a function is used to
+#' reset a parameter value, the \code{nrounds} argument in this function would be the
 #' the number of boosting rounds in the current training.
 #'
 #' Callback function expects the following values to be set in its calling frame:
@@ -176,32 +176,32 @@ cb.evaluation.log <- function() {
 #' \code{iteration},
 #' \code{begin_iteration},
 #' \code{end_iteration}.
-#' 
+#'
 #' @seealso
 #' \code{\link{callbacks}}
-#' 
+#'
 #' @export
 cb.reset.parameters <- function(new_params) {
 
-  if (typeof(new_params) != "list") 
+  if (typeof(new_params) != "list")
     stop("'new_params' must be a list")
   pnames <- gsub("\\.", "_", names(new_params))
   nrounds <- NULL
-  
+
   # run some checks in the begining
   init <- function(env) {
     nrounds <<- env$end_iteration - env$begin_iteration + 1
-    
+
     if (is.null(env$bst) && is.null(env$bst_folds))
       stop("Parent frame has neither 'bst' nor 'bst_folds'")
-    
+
     # Some parameters are not allowed to be changed,
     # since changing them would simply wreck some chaos
-    not_allowed <- pnames %in% 
+    not_allowed <- pnames %in%
       c('num_class', 'num_output_group', 'size_leaf_vector', 'updater_seq')
     if (any(not_allowed))
       stop('Parameters ', paste(pnames[not_allowed]), " cannot be changed during boosting.")
-    
+
     for (n in pnames) {
       p <- new_params[[n]]
       if (is.function(p)) {
@@ -215,18 +215,18 @@ cb.reset.parameters <- function(new_params) {
       }
     }
   }
-  
+
   callback <- function(env = parent.frame()) {
     if (is.null(nrounds))
       init(env)
-    
+
     i <- env$iteration
     pars <- lapply(new_params, function(p) {
       if (is.function(p))
         return(p(i, nrounds))
       p[i]
     })
-    
+
     if (!is.null(env$bst)) {
       xgb.parameters(env$bst$handle) <- pars
     } else {
@@ -242,23 +242,23 @@ cb.reset.parameters <- function(new_params) {
 
 
 #' Callback closure to activate the early stopping.
-#' 
-#' @param stopping_rounds The number of rounds with no improvement in 
+#'
+#' @param stopping_rounds The number of rounds with no improvement in
 #'        the evaluation metric in order to stop the training.
 #' @param maximize whether to maximize the evaluation metric
 #' @param metric_name the name of an evaluation column to use as a criteria for early
 #'        stopping. If not set, the last column would be used.
-#'        Let's say the test data in \code{watchlist} was labelled as \code{dtest}, 
-#'        and one wants to use the AUC in test data for early stopping regardless of where 
+#'        Let's say the test data in \code{watchlist} was labelled as \code{dtest},
+#'        and one wants to use the AUC in test data for early stopping regardless of where
 #'        it is in the \code{watchlist}, then one of the following would need to be set:
 #'        \code{metric_name='dtest-auc'} or \code{metric_name='dtest_auc'}.
 #'        All dash '-' characters in metric names are considered equivalent to '_'.
 #' @param verbose whether to print the early stopping information.
-#' 
+#'
 #' @details
-#' This callback function determines the condition for early stopping 
+#' This callback function determines the condition for early stopping
 #' by setting the \code{stop_condition = TRUE} flag in its calling frame.
-#' 
+#'
 #' The following additional fields are assigned to the model's R object:
 #' \itemize{
 #' \item \code{best_score} the evaluation score at the best iteration
@@ -266,13 +266,13 @@ cb.reset.parameters <- function(new_params) {
 #' \item \code{best_ntreelimit} to use with the \code{ntreelimit} parameter in \code{predict}.
 #'      It differs from \code{best_iteration} in multiclass or random forest settings.
 #' }
-#' 
+#'
 #' The Same values are also stored as xgb-attributes:
 #' \itemize{
 #' \item \code{best_iteration} is stored as a 0-based iteration index (for interoperability of binary models)
 #' \item \code{best_msg} message string is also stored.
 #' }
-#' 
+#'
 #' At least one data element is required in the evaluation watchlist for early stopping to work.
 #'
 #' Callback function expects the following values to be set in its calling frame:
@@ -284,13 +284,13 @@ cb.reset.parameters <- function(new_params) {
 #' \code{begin_iteration},
 #' \code{end_iteration},
 #' \code{num_parallel_tree}.
-#' 
+#'
 #' @seealso
 #' \code{\link{callbacks}},
 #' \code{\link{xgb.attr}}
-#' 
+#'
 #' @export
-cb.early.stop <- function(stopping_rounds, maximize = FALSE, 
+cb.early.stop <- function(stopping_rounds, maximize = FALSE,
                           metric_name = NULL, verbose = TRUE) {
   # state variables
   best_iteration <- -1
@@ -298,11 +298,11 @@ cb.early.stop <- function(stopping_rounds, maximize = FALSE,
   best_score <- Inf
   best_msg <- NULL
   metric_idx <- 1
-  
+
   init <- function(env) {
     if (length(env$bst_evaluation) == 0)
       stop("For early stopping, watchlist must have at least one element")
-    
+
     eval_names <- gsub('-', '_', names(env$bst_evaluation))
     if (!is.null(metric_name)) {
       metric_idx <<- which(gsub('-', '_', metric_name) == eval_names)
@@ -314,25 +314,25 @@ cb.early.stop <- function(stopping_rounds, maximize = FALSE,
         length(env$bst_evaluation) > 1) {
       metric_idx <<- length(eval_names)
       if (verbose)
-        cat('Multiple eval metrics are present. Will use ', 
+        cat('Multiple eval metrics are present. Will use ',
             eval_names[metric_idx], ' for early stopping.\n', sep = '')
     }
-    
+
     metric_name <<- eval_names[metric_idx]
-    
+
     # maximize is usually NULL when not set in xgb.train and built-in metrics
     if (is.null(maximize))
       maximize <<- grepl('(_auc|_map|_ndcg)', metric_name)
 
     if (verbose && NVL(env$rank, 0) == 0)
-      cat("Will train until ", metric_name, " hasn't improved in ", 
+      cat("Will train until ", metric_name, " hasn't improved in ",
           stopping_rounds, " rounds.\n\n", sep = '')
 
     best_iteration <<- 1
     if (maximize) best_score <<- -Inf
-    
+
     env$stop_condition <- FALSE
-    
+
     if (!is.null(env$bst)) {
       if (!inherits(env$bst, 'xgb.Booster'))
         stop("'bst' in the parent frame must be an 'xgb.Booster'")
@@ -348,7 +348,7 @@ cb.early.stop <- function(stopping_rounds, maximize = FALSE,
       stop("Parent frame has neither 'bst' nor ('bst_folds' and 'basket')")
     }
   }
-  
+
   finalizer <- function(env) {
     if (!is.null(env$bst)) {
       attr_best_score = as.numeric(xgb.attr(env$bst$handle, 'best_score'))
@@ -367,16 +367,16 @@ cb.early.stop <- function(stopping_rounds, maximize = FALSE,
   callback <- function(env = parent.frame(), finalize = FALSE) {
     if (best_iteration < 0)
       init(env)
-    
+
     if (finalize)
       return(finalizer(env))
-    
+
     i <- env$iteration
     score = env$bst_evaluation[metric_idx]
-    
+
     if (( maximize && score > best_score) ||
         (!maximize && score < best_score)) {
-      
+
       best_msg <<- format.eval.string(i, env$bst_evaluation, env$bst_evaluation_err)
       best_score <<- score
       best_iteration <<- i
@@ -403,37 +403,37 @@ cb.early.stop <- function(stopping_rounds, maximize = FALSE,
 
 
 #' Callback closure for saving a model file.
-#' 
-#' @param save_period save the model to disk after every 
+#'
+#' @param save_period save the model to disk after every
 #'        \code{save_period} iterations; 0 means save the model at the end.
 #' @param save_name the name or path for the saved model file.
-#'        It can contain a \code{\link[base]{sprintf}} formatting specifier 
+#'        It can contain a \code{\link[base]{sprintf}} formatting specifier
 #'        to include the integer iteration number in the file name.
-#'        E.g., with \code{save_name} = 'xgboost_%04d.model', 
+#'        E.g., with \code{save_name} = 'xgboost_%04d.model',
 #'        the file saved at iteration 50 would be named "xgboost_0050.model".
-#' 
-#' @details 
+#'
+#' @details
 #' This callback function allows to save an xgb-model file, either periodically after each \code{save_period}'s or at the end.
-#' 
+#'
 #' Callback function expects the following values to be set in its calling frame:
 #' \code{bst},
 #' \code{iteration},
 #' \code{begin_iteration},
 #' \code{end_iteration}.
-#' 
+#'
 #' @seealso
 #' \code{\link{callbacks}}
-#' 
+#'
 #' @export
 cb.save.model <- function(save_period = 0, save_name = "xgboost.model") {
-  
+
   if (save_period < 0)
     stop("'save_period' cannot be negative")
 
   callback <- function(env = parent.frame()) {
     if (is.null(env$bst))
       stop("'save_model' callback requires the 'bst' booster object in its calling frame")
-    
+
     if ((save_period > 0 && (env$iteration - env$begin_iteration) %% save_period == 0) ||
         (save_period == 0 && env$iteration == env$end_iteration))
       xgb.save(env$bst, sprintf(save_name, env$iteration))
@@ -445,16 +445,16 @@ cb.save.model <- function(save_period = 0, save_name = "xgboost.model") {
 
 
 #' Callback closure for returning cross-validation based predictions.
-#' 
+#'
 #' @param save_models a flag for whether to save the folds' models.
-#' 
-#' @details 
+#'
+#' @details
 #' This callback function saves predictions for all of the test folds,
 #' and also allows to save the folds' models.
-#' 
+#'
 #' It is a "finalizer" callback and it uses early stopping information whenever it is available,
 #' thus it must be run after the early stopping callback if the early stopping is used.
-#' 
+#'
 #' Callback function expects the following values to be set in its calling frame:
 #' \code{bst_folds},
 #' \code{basket},
@@ -463,36 +463,36 @@ cb.save.model <- function(save_period = 0, save_name = "xgboost.model") {
 #' \code{params},
 #' \code{num_parallel_tree},
 #' \code{num_class}.
-#' 
-#' @return 
+#'
+#' @return
 #' Predictions are returned inside of the \code{pred} element, which is either a vector or a matrix,
-#' depending on the number of prediction outputs per data row. The order of predictions corresponds 
-#' to the order of rows in the original dataset. Note that when a custom \code{folds} list is 
-#' provided in \code{xgb.cv}, the predictions would only be returned properly when this list is a 
-#' non-overlapping list of k sets of indices, as in a standard k-fold CV. The predictions would not be 
-#' meaningful when user-profided folds have overlapping indices as in, e.g., random sampling splits.
+#' depending on the number of prediction outputs per data row. The order of predictions corresponds
+#' to the order of rows in the original dataset. Note that when a custom \code{folds} list is
+#' provided in \code{xgb.cv}, the predictions would only be returned properly when this list is a
+#' non-overlapping list of k sets of indices, as in a standard k-fold CV. The predictions would not be
+#' meaningful when user-provided folds have overlapping indices as in, e.g., random sampling splits.
 #' When some of the indices in the training dataset are not included into user-provided \code{folds},
 #' their prediction value would be \code{NA}.
-#' 
+#'
 #' @seealso
 #' \code{\link{callbacks}}
-#' 
+#'
 #' @export
 cb.cv.predict <- function(save_models = FALSE) {
-  
+
   finalizer <- function(env) {
     if (is.null(env$basket) || is.null(env$bst_folds))
       stop("'cb.cv.predict' callback requires 'basket' and 'bst_folds' lists in its calling frame")
-    
+
     N <- nrow(env$data)
-    pred <- 
+    pred <-
       if (env$num_class > 1) {
         matrix(NA_real_, N, env$num_class)
       } else {
         rep(NA_real_, N)
       }
 
-    ntreelimit <- NVL(env$basket$best_ntreelimit, 
+    ntreelimit <- NVL(env$basket$best_ntreelimit,
                       env$end_iteration * env$num_parallel_tree)
     if (NVL(env$params[['booster']], '') == 'gblinear') {
       ntreelimit <- 0 # must be 0 for gblinear
@@ -569,7 +569,7 @@ cb.cv.predict <- function(save_models = FALSE) {
 #' # Extract the coefficients' path and plot them vs boosting iteration number:
 #' coef_path <- xgb.gblinear.history(bst)
 #' matplot(coef_path, type = 'l')
-#' 
+#'
 #' # With the deterministic coordinate descent updater, it is safer to use higher learning rates.
 #' # Will try the classical componentwise boosting which selects a single best feature per round:
 #' bst <- xgb.train(param, dtrain, list(tr=dtrain), nrounds = 200, eta = 0.8,
@@ -586,7 +586,7 @@ cb.cv.predict <- function(save_models = FALSE) {
 #' # coefficients in the CV fold #3
 #' xgb.gblinear.history(bst)[[3]] %>% matplot(type = 'l')
 #'
-#' 
+#'
 #' #### Multiclass classification:
 #' #
 #' dtrain <- xgb.DMatrix(scale(x), label = as.numeric(iris$Species) - 1)
@@ -681,9 +681,9 @@ cb.gblinear.history <- function(sparse=FALSE) {
 #'        using the \code{cb.gblinear.history()} callback.
 #' @param class_index zero-based class index to extract the coefficients for only that
 #'        specific class in a multinomial multiclass model. When it is NULL, all the
-#'        coeffients are returned. Has no effect in non-multiclass models.
+#'        coefficients are returned. Has no effect in non-multiclass models.
 #'
-#' @return 
+#' @return
 #' For an \code{xgb.train} result, a matrix (either dense or sparse) with the columns
 #' corresponding to iteration's coefficients (in the order as \code{xgb.dump()} would
 #' return) and the rows corresponding to boosting iterations.
@@ -731,7 +731,7 @@ xgb.gblinear.history <- function(model, class_index = NULL) {
   coef_path <- environment(model$callbacks$cb.gblinear.history)[["coefs"]]
   if (!is.null(class_index) && num_class > 1) {
     coef_path <- if (is.list(coef_path)) {
-      lapply(coef_path, 
+      lapply(coef_path,
              function(x) x[, seq(1 + class_index, by=num_class, length.out=num_feat)])
     } else {
       coef_path <- coef_path[, seq(1 + class_index, by=num_class, length.out=num_feat)]
@@ -743,7 +743,7 @@ xgb.gblinear.history <- function(model, class_index = NULL) {
 
 #
 # Internal utility functions for callbacks ------------------------------------
-# 
+#
 
 # Format the evaluation metric string
 format.eval.string <- function(iter, eval_res, eval_err = NULL) {
@@ -773,7 +773,7 @@ callback.calls <- function(cb_list) {
   unlist(lapply(cb_list, function(x) attr(x, 'call')))
 }
 
-# Add a callback cb to the list and make sure that 
+# Add a callback cb to the list and make sure that
 # cb.early.stop and cb.cv.predict are at the end of the list
 # with cb.cv.predict being the last (when present)
 add.cb <- function(cb_list, cb) {
@@ -782,11 +782,11 @@ add.cb <- function(cb_list, cb) {
   if ('cb.early.stop' %in% names(cb_list)) {
     cb_list <- c(cb_list, cb_list['cb.early.stop'])
     # this removes only the first one
-    cb_list['cb.early.stop'] <- NULL 
+    cb_list['cb.early.stop'] <- NULL
   }
   if ('cb.cv.predict' %in% names(cb_list)) {
     cb_list <- c(cb_list, cb_list['cb.cv.predict'])
-    cb_list['cb.cv.predict'] <- NULL 
+    cb_list['cb.cv.predict'] <- NULL
   }
   cb_list
 }
@@ -796,7 +796,7 @@ categorize.callbacks <- function(cb_list) {
   list(
     pre_iter = Filter(function(x) {
         pre <- attr(x, 'is_pre_iteration')
-        !is.null(pre) && pre 
+        !is.null(pre) && pre
       }, cb_list),
     post_iter = Filter(function(x) {
         pre <- attr(x, 'is_pre_iteration')

R-package/R/utils.R

@@ -28,12 +28,12 @@ NVL <- function(x, val) {
 # Merges booster params with whatever is provided in ...
 # plus runs some checks
 check.booster.params <- function(params, ...) {
-  if (typeof(params) != "list") 
+  if (typeof(params) != "list")
     stop("params must be a list")
-  
+
   # in R interface, allow for '.' instead of '_' in parameter names
   names(params) <- gsub("\\.", "_", names(params))
-  
+
   # merge parameters from the params and the dots-expansion
   dot_params <- list(...)
   names(dot_params) <- gsub("\\.", "_", names(dot_params))
@@ -41,15 +41,15 @@ check.booster.params <- function(params, ...) {
                        names(dot_params))) > 0)
     stop("Same parameters in 'params' and in the call are not allowed. Please check your 'params' list.")
   params <- c(params, dot_params)
-  
+
   # providing a parameter multiple times makes sense only for 'eval_metric'
   name_freqs <- table(names(params))
   multi_names <- setdiff(names(name_freqs[name_freqs > 1]), 'eval_metric')
   if (length(multi_names) > 0) {
     warning("The following parameters were provided multiple times:\n\t",
             paste(multi_names, collapse = ', '), "\n  Only the last value for each of them will be used.\n")
-    # While xgboost internals would choose the last value for a multiple-times parameter, 
-    # enforce it here in R as well (b/c multi-parameters might be used further in R code, 
+    # While xgboost internals would choose the last value for a multiple-times parameter,
+    # enforce it here in R as well (b/c multi-parameters might be used further in R code,
     # and R takes the 1st value when multiple elements with the same name are present in a list).
     for (n in multi_names) {
       del_idx <- which(n == names(params))
@@ -57,23 +57,36 @@ check.booster.params <- function(params, ...) {
       params[[del_idx]] <- NULL
     }
   }
-  
+
   # for multiclass, expect num_class to be set
   if (typeof(params[['objective']]) == "character" &&
       substr(NVL(params[['objective']], 'x'), 1, 6) == 'multi:' &&
       as.numeric(NVL(params[['num_class']], 0)) < 2) {
         stop("'num_class' > 1 parameter must be set for multiclass classification")
   }
-  
+
   # monotone_constraints parser
-  
+
   if (!is.null(params[['monotone_constraints']]) &&
       typeof(params[['monotone_constraints']]) != "character") {
         vec2str = paste(params[['monotone_constraints']], collapse = ',')
         vec2str = paste0('(', vec2str, ')')
         params[['monotone_constraints']] = vec2str
   }
-  
+
+  # interaction constraints parser (convert from list of column indices to string)
+  if (!is.null(params[['interaction_constraints']]) &&
+      typeof(params[['interaction_constraints']]) != "character"){
+    # check input class
+    if (class(params[['interaction_constraints']]) != 'list') stop('interaction_constraints should be class list')
+    if (!all(unique(sapply(params[['interaction_constraints']], class)) %in% c('numeric','integer'))) {
+      stop('interaction_constraints should be a list of numeric/integer vectors')
+    }
+
+    # recast parameter as string
+    interaction_constraints <- sapply(params[['interaction_constraints']], function(x) paste0('[', paste(x, collapse=','), ']'))
+    params[['interaction_constraints']] <- paste0('[', paste(interaction_constraints, collapse=','), ']')
+  }
   return(params)
 }
 
@@ -83,10 +96,10 @@ check.booster.params <- function(params, ...) {
 check.custom.obj <- function(env = parent.frame()) {
   if (!is.null(env$params[['objective']]) && !is.null(env$obj))
     stop("Setting objectives in 'params' and 'obj' at the same time is not allowed")
-  
+
   if (!is.null(env$obj) && typeof(env$obj) != 'closure')
     stop("'obj' must be a function")
-  
+
   # handle the case when custom objective function was provided through params
   if (!is.null(env$params[['objective']]) &&
       typeof(env$params$objective) == 'closure') {
@@ -100,21 +113,21 @@ check.custom.obj <- function(env = parent.frame()) {
 check.custom.eval <- function(env = parent.frame()) {
   if (!is.null(env$params[['eval_metric']]) && !is.null(env$feval))
     stop("Setting evaluation metrics in 'params' and 'feval' at the same time is not allowed")
-  
+
   if (!is.null(env$feval) && typeof(env$feval) != 'closure')
     stop("'feval' must be a function")
-  
+
   # handle a situation when custom eval function was provided through params
   if (!is.null(env$params[['eval_metric']]) &&
       typeof(env$params$eval_metric) == 'closure') {
     env$feval <- env$params$eval_metric
     env$params$eval_metric <- NULL
   }
-  
+
   # require maximize to be set when custom feval and early stopping are used together
   if (!is.null(env$feval) &&
       is.null(env$maximize) && (
-        !is.null(env$early_stopping_rounds) || 
+        !is.null(env$early_stopping_rounds) ||
         has.callbacks(env$callbacks, 'cb.early.stop')))
     stop("Please set 'maximize' to indicate whether the evaluation metric needs to be maximized or not")
 }
@@ -141,15 +154,15 @@ xgb.iter.update <- function(booster_handle, dtrain, iter, obj = NULL) {
 
 
 # Evaluate one iteration.
-# Returns a named vector of evaluation metrics 
+# Returns a named vector of evaluation metrics
 # with the names in a 'datasetname-metricname' format.
 xgb.iter.eval <- function(booster_handle, watchlist, iter, feval = NULL) {
   if (!identical(class(booster_handle), "xgb.Booster.handle"))
     stop("class of booster_handle must be xgb.Booster.handle")
 
-  if (length(watchlist) == 0) 
+  if (length(watchlist) == 0)
     return(NULL)
-  
+
   evnames <- names(watchlist)
   if (is.null(feval)) {
     msg <- .Call(XGBoosterEvalOneIter_R, booster_handle, as.integer(iter), watchlist, as.list(evnames))
@@ -176,7 +189,7 @@ xgb.iter.eval <- function(booster_handle, watchlist, iter, feval = NULL) {
 
 # Generates random (stratified if needed) CV folds
 generate.cv.folds <- function(nfold, nrows, stratified, label, params) {
-  
+
   # cannot do it for rank
   if (exists('objective', where = params) &&
       is.character(params$objective) &&
@@ -196,13 +209,14 @@ generate.cv.folds <- function(nfold, nrows, stratified, label, params) {
     if (exists('objective', where = params) &&
         is.character(params$objective)) {
       # If 'objective' provided in params, assume that y is a classification label
-      # unless objective is reg:linear
-      if (params$objective != 'reg:linear')
+      # unless objective is reg:squarederror
+      if (params$objective != 'reg:squarederror')
         y <- factor(y)
     } else {
-      # If no 'objective' given in params, it means that user either wants to use
-      # the default 'reg:linear' objective or has provided a custom obj function.
-      # Here, assume classification setting when y has 5 or less unique values:
+      # If no 'objective' given in params, it means that user either wants to
+      # use the default 'reg:squarederror' objective or has provided a custom
+      # obj function.  Here, assume classification setting when y has 5 or less
+      # unique values:
       if (length(unique(y)) <= 5)
         y <- factor(y)
     }
@@ -262,7 +276,8 @@ xgb.createFolds <- function(y, k = 10)
       ## add enough random integers to get  length(seqVector) == numInClass[i]
       if (numInClass[i] %% k > 0) seqVector <- c(seqVector, sample.int(k, numInClass[i] %% k))
       ## shuffle the integers for fold assignment and assign to this classes's data
-      foldVector[y == dimnames(numInClass)$y[i]] <- sample(seqVector)
+      ## seqVector[sample.int(length(seqVector))] is used to handle length(seqVector) == 1
+      foldVector[y == dimnames(numInClass)$y[i]] <- seqVector[sample.int(length(seqVector))]
     }
   } else {
     foldVector <- seq(along = y)
@@ -279,22 +294,22 @@ xgb.createFolds <- function(y, k = 10)
 #
 
 #' Deprecation notices.
-#' 
+#'
 #' At this time, some of the parameter names were changed in order to make the code style more uniform.
 #' The deprecated parameters would be removed in the next release.
-#' 
+#'
 #' To see all the current deprecated and new parameters, check the \code{xgboost:::depr_par_lut} table.
-#' 
-#' A deprecation warning is shown when any of the deprecated parameters is used in a call. 
-#' An additional warning is shown when there was a partial match to a deprecated parameter 
+#'
+#' A deprecation warning is shown when any of the deprecated parameters is used in a call.
+#' An additional warning is shown when there was a partial match to a deprecated parameter
 #' (as R is able to partially match parameter names).
-#' 
+#'
 #' @name xgboost-deprecated
 NULL
 
 # Lookup table for the deprecated parameters bookkeeping
 depr_par_lut <- matrix(c(
-  'print.every.n', 'print_every_n', 
+  'print.every.n', 'print_every_n',
   'early.stop.round', 'early_stopping_rounds',
   'training.data', 'data',
   'with.stats', 'with_stats',

R-package/R/xgb.Booster.R

@@ -81,7 +81,7 @@ xgb.get.handle <- function(object) {
 #' its handle (pointer) to an internal xgboost model would be invalid. The majority of xgboost methods
 #' should still work for such a model object since those methods would be using
 #' \code{xgb.Booster.complete} internally. However, one might find it to be more efficient to call the
-#' \code{xgb.Booster.complete} function explicitely once after loading a model as an R-object.
+#' \code{xgb.Booster.complete} function explicitly once after loading a model as an R-object.
 #' That would prevent further repeated implicit reconstruction of an internal booster model.
 #'
 #' @return
@@ -95,6 +95,7 @@ xgb.get.handle <- function(object) {
 #' saveRDS(bst, "xgb.model.rds")
 #'
 #' bst1 <- readRDS("xgb.model.rds")
+#' if (file.exists("xgb.model.rds")) file.remove("xgb.model.rds")
 #' # the handle is invalid:
 #' print(bst1$handle)
 #'
@@ -129,11 +130,13 @@ xgb.Booster.complete <- function(object, saveraw = TRUE) {
 #'        logistic regression would result in predictions for log-odds instead of probabilities.
 #' @param ntreelimit limit the number of model's trees or boosting iterations used in prediction (see Details).
 #'        It will use all the trees by default (\code{NULL} value).
-#' @param predleaf whether predict leaf index instead.
-#' @param predcontrib whether to return feature contributions to individual predictions instead (see Details).
+#' @param predleaf whether predict leaf index.
+#' @param predcontrib whether to return feature contributions to individual predictions (see Details).
 #' @param approxcontrib whether to use a fast approximation for feature contributions (see Details).
+#' @param predinteraction whether to return contributions of feature interactions to individual predictions (see Details).
 #' @param reshape whether to reshape the vector of predictions to a matrix form when there are several
-#'        prediction outputs per case. This option has no effect when \code{predleaf = TRUE}.
+#'        prediction outputs per case. This option has no effect when either of predleaf, predcontrib,
+#'        or predinteraction flags is TRUE.
 #' @param ... Parameters passed to \code{predict.xgb.Booster}
 #'
 #' @details
@@ -158,6 +161,11 @@ xgb.Booster.complete <- function(object, saveraw = TRUE) {
 #' Setting \code{approxcontrib = TRUE} approximates these values following the idea explained
 #' in \url{http://blog.datadive.net/interpreting-random-forests/}.
 #'
+#' With \code{predinteraction = TRUE}, SHAP values of contributions of interaction of each pair of features
+#' are computed. Note that this operation might be rather expensive in terms of compute and memory.
+#' Since it quadratically depends on the number of features, it is recommended to perform selection
+#' of the most important features first. See below about the format of the returned results.
+#'
 #' @return
 #' For regression or binary classification, it returns a vector of length \code{nrows(newdata)}.
 #' For multiclass classification, either a \code{num_class * nrows(newdata)} vector or
@@ -173,9 +181,17 @@ xgb.Booster.complete <- function(object, saveraw = TRUE) {
 #' such a matrix. The contribution values are on the scale of untransformed margin
 #' (e.g., for binary classification would mean that the contributions are log-odds deviations from bias).
 #'
+#' When \code{predinteraction = TRUE} and it is not a multiclass setting, the output is a 3d array with
+#' dimensions \code{c(nrow, num_features + 1, num_features + 1)}. The off-diagonal (in the last two dimensions)
+#' elements represent different features interaction contributions. The array is symmetric WRT the last
+#' two dimensions. The "+ 1" columns corresponds to bias. Summing this array along the last dimension should
+#' produce practically the same result as predict with \code{predcontrib = TRUE}.
+#' For a multiclass case, a list of \code{num_class} elements is returned, where each element is
+#' such an array.
+#'
 #' @seealso
 #' \code{\link{xgb.train}}.
-#' 
+#'
 #' @references
 #'
 #' Scott M. Lundberg, Su-In Lee, "A Unified Approach to Interpreting Model Predictions", NIPS Proceedings 2017, \url{https://arxiv.org/abs/1705.07874}
@@ -269,7 +285,8 @@ xgb.Booster.complete <- function(object, saveraw = TRUE) {
 #' @rdname predict.xgb.Booster
 #' @export
 predict.xgb.Booster <- function(object, newdata, missing = NA, outputmargin = FALSE, ntreelimit = NULL,
-                                predleaf = FALSE, predcontrib = FALSE, approxcontrib = FALSE, reshape = FALSE, ...) {
+                                predleaf = FALSE, predcontrib = FALSE, approxcontrib = FALSE, predinteraction = FALSE,
+                                reshape = FALSE, ...) {
 
   object <- xgb.Booster.complete(object, saveraw = FALSE)
   if (!inherits(newdata, "xgb.DMatrix"))
@@ -285,7 +302,8 @@ predict.xgb.Booster <- function(object, newdata, missing = NA, outputmargin = FA
   if (ntreelimit < 0)
     stop("ntreelimit cannot be negative")
 
-  option <- 0L + 1L * as.logical(outputmargin) + 2L * as.logical(predleaf) + 4L * as.logical(predcontrib) + 8L * as.logical(approxcontrib)
+  option <- 0L + 1L * as.logical(outputmargin) + 2L * as.logical(predleaf) + 4L * as.logical(predcontrib) +
+    8L * as.logical(approxcontrib) + 16L * as.logical(predinteraction)
 
   ret <- .Call(XGBoosterPredict_R, object$handle, newdata, option[1], as.integer(ntreelimit))
 
@@ -305,17 +323,28 @@ predict.xgb.Booster <- function(object, newdata, missing = NA, outputmargin = FA
   } else if (predcontrib) {
     n_col1 <- ncol(newdata) + 1
     n_group <- npred_per_case / n_col1
-    dnames <- if (!is.null(colnames(newdata))) list(NULL, c(colnames(newdata), "BIAS")) else NULL
+    cnames <- if (!is.null(colnames(newdata))) c(colnames(newdata), "BIAS") else NULL
     ret <- if (n_ret == n_row) {
-      matrix(ret, ncol = 1, dimnames = dnames)
+      matrix(ret, ncol = 1, dimnames = list(NULL, cnames))
     } else if (n_group == 1) {
-      matrix(ret, nrow = n_row, byrow = TRUE, dimnames = dnames)
+      matrix(ret, nrow = n_row, byrow = TRUE, dimnames = list(NULL, cnames))
     } else {
-      grp_mask <- rep(seq_len(n_col1), n_row) +
-        rep((seq_len(n_row) - 1) * n_col1 * n_group, each = n_col1)
-      lapply(seq_len(n_group), function(g) {
-        matrix(ret[grp_mask + n_col1 * (g - 1)], nrow = n_row, byrow = TRUE, dimnames = dnames)
-      })
+      arr <- array(ret, c(n_col1, n_group, n_row),
+                   dimnames = list(cnames, NULL, NULL)) %>% aperm(c(2,3,1)) # [group, row, col]
+      lapply(seq_len(n_group), function(g) arr[g,,])
+    }
+  } else if (predinteraction) {
+    n_col1 <- ncol(newdata) + 1
+    n_group <- npred_per_case / n_col1^2
+    cnames <- if (!is.null(colnames(newdata))) c(colnames(newdata), "BIAS") else NULL
+    ret <- if (n_ret == n_row) {
+      matrix(ret, ncol = 1, dimnames = list(NULL, cnames))
+    } else if (n_group == 1) {
+      array(ret, c(n_col1, n_col1, n_row), dimnames = list(cnames, cnames, NULL)) %>% aperm(c(3,1,2))
+    } else {
+      arr <- array(ret, c(n_col1, n_col1, n_group, n_row),
+                   dimnames = list(cnames, cnames, NULL, NULL)) %>% aperm(c(3,4,1,2)) # [group, row, col1, col2]
+      lapply(seq_len(n_group), function(g) arr[g,,,])
     }
   } else if (reshape && npred_per_case > 1) {
     ret <- matrix(ret, nrow = n_row, byrow = TRUE)
@@ -390,6 +419,7 @@ predict.xgb.Booster.handle <- function(object, ...) {
 #'
 #' xgb.save(bst, 'xgb.model')
 #' bst1 <- xgb.load('xgb.model')
+#' if (file.exists('xgb.model')) file.remove('xgb.model')
 #' print(xgb.attr(bst1, "my_attribute"))
 #' print(xgb.attributes(bst1))
 #'

R-package/R/xgb.DMatrix.R

@@ -1,24 +1,25 @@
 #' Construct xgb.DMatrix object
-#' 
+#'
 #' Construct xgb.DMatrix object from either a dense matrix, a sparse matrix, or a local file.
 #' Supported input file formats are either a libsvm text file or a binary file that was created previously by
 #' \code{\link{xgb.DMatrix.save}}).
-#' 
-#' @param data a \code{matrix} object (either numeric or integer), a \code{dgCMatrix} object, or a character 
+#'
+#' @param data a \code{matrix} object (either numeric or integer), a \code{dgCMatrix} object, or a character
 #'        string representing a filename.
 #' @param info a named list of additional information to store in the \code{xgb.DMatrix} object.
-#'        See \code{\link{setinfo}} for the specific allowed kinds of 
+#'        See \code{\link{setinfo}} for the specific allowed kinds of
 #' @param missing a float value to represents missing values in data (used only when input is a dense matrix).
 #'        It is useful when a 0 or some other extreme value represents missing values in data.
 #' @param silent whether to suppress printing an informational message after loading from a file.
 #' @param ... the \code{info} data could be passed directly as parameters, without creating an \code{info} list.
-#' 
+#'
 #' @examples
 #' data(agaricus.train, package='xgboost')
 #' train <- agaricus.train
 #' dtrain <- xgb.DMatrix(train$data, label=train$label)
 #' xgb.DMatrix.save(dtrain, 'xgb.DMatrix.data')
 #' dtrain <- xgb.DMatrix('xgb.DMatrix.data')
+#' if (file.exists('xgb.DMatrix.data')) file.remove('xgb.DMatrix.data')
 #' @export
 xgb.DMatrix <- function(data, info = list(), missing = NA, silent = FALSE, ...) {
   cnames <- NULL
@@ -78,23 +79,23 @@ xgb.get.DMatrix <- function(data, label = NULL, missing = NA, weight = NULL) {
 
 
 #' Dimensions of xgb.DMatrix
-#' 
+#'
 #' Returns a vector of numbers of rows and of columns in an \code{xgb.DMatrix}.
 #' @param x Object of class \code{xgb.DMatrix}
-#' 
+#'
 #' @details
-#' Note: since \code{nrow} and \code{ncol} internally use \code{dim}, they can also 
+#' Note: since \code{nrow} and \code{ncol} internally use \code{dim}, they can also
 #' be directly used with an \code{xgb.DMatrix} object.
 #'
 #' @examples
 #' data(agaricus.train, package='xgboost')
 #' train <- agaricus.train
 #' dtrain <- xgb.DMatrix(train$data, label=train$label)
-#' 
+#'
 #' stopifnot(nrow(dtrain) == nrow(train$data))
 #' stopifnot(ncol(dtrain) == ncol(train$data))
 #' stopifnot(all(dim(dtrain) == dim(train$data)))
-#' 
+#'
 #' @export
 dim.xgb.DMatrix <- function(x) {
   c(.Call(XGDMatrixNumRow_R, x), .Call(XGDMatrixNumCol_R, x))
@@ -102,14 +103,14 @@ dim.xgb.DMatrix <- function(x) {
 
 
 #' Handling of column names of \code{xgb.DMatrix}
-#' 
-#' Only column names are supported for \code{xgb.DMatrix}, thus setting of 
-#' row names would have no effect and returnten row names would be NULL.
-#' 
+#'
+#' Only column names are supported for \code{xgb.DMatrix}, thus setting of
+#' row names would have no effect and returned row names would be NULL.
+#'
 #' @param x object of class \code{xgb.DMatrix}
 #' @param value a list of two elements: the first one is ignored
-#'        and the second one is column names 
-#' 
+#'        and the second one is column names
+#'
 #' @details
 #' Generic \code{dimnames} methods are used by \code{colnames}.
 #' Since row names are irrelevant, it is recommended to use \code{colnames} directly.
@@ -122,7 +123,7 @@ dim.xgb.DMatrix <- function(x) {
 #' colnames(dtrain)
 #' colnames(dtrain) <- make.names(1:ncol(train$data))
 #' print(dtrain, verbose=TRUE)
-#' 
+#'
 #' @rdname dimnames.xgb.DMatrix
 #' @export
 dimnames.xgb.DMatrix <- function(x) {
@@ -140,8 +141,8 @@ dimnames.xgb.DMatrix <- function(x) {
     attr(x, '.Dimnames') <- NULL
     return(x)
   }
-  if (ncol(x) != length(value[[2]])) 
-    stop("can't assign ", length(value[[2]]), " colnames to a ", 
+  if (ncol(x) != length(value[[2]]))
+    stop("can't assign ", length(value[[2]]), " colnames to a ",
          ncol(x), " column xgb.DMatrix")
   attr(x, '.Dimnames') <- value
   x
@@ -149,33 +150,33 @@ dimnames.xgb.DMatrix <- function(x) {
 
 
 #' Get information of an xgb.DMatrix object
-#' 
+#'
 #' Get information of an xgb.DMatrix object
 #' @param object Object of class \code{xgb.DMatrix}
 #' @param name the name of the information field to get (see details)
 #' @param ... other parameters
-#' 
+#'
 #' @details
 #' The \code{name} field can be one of the following:
-#' 
+#'
 #' \itemize{
 #'     \item \code{label}: label Xgboost learn from ;
 #'     \item \code{weight}: to do a weight rescale ;
 #'     \item \code{base_margin}: base margin is the base prediction Xgboost will boost from ;
 #'     \item \code{nrow}: number of rows of the \code{xgb.DMatrix}.
-#'     
+#'
 #' }
-#' 
+#'
 #' \code{group} can be setup by \code{setinfo} but can't be retrieved by \code{getinfo}.
-#' 
+#'
 #' @examples
 #' data(agaricus.train, package='xgboost')
 #' train <- agaricus.train
 #' dtrain <- xgb.DMatrix(train$data, label=train$label)
-#' 
+#'
 #' labels <- getinfo(dtrain, 'label')
 #' setinfo(dtrain, 'label', 1-labels)
-#' 
+#'
 #' labels2 <- getinfo(dtrain, 'label')
 #' stopifnot(all(labels2 == 1-labels))
 #' @rdname getinfo
@@ -202,9 +203,9 @@ getinfo.xgb.DMatrix <- function(object, name, ...) {
 
 
 #' Set information of an xgb.DMatrix object
-#' 
+#'
 #' Set information of an xgb.DMatrix object
-#' 
+#'
 #' @param object Object of class "xgb.DMatrix"
 #' @param name the name of the field to get
 #' @param info the specific field of information to set
@@ -212,19 +213,19 @@ getinfo.xgb.DMatrix <- function(object, name, ...) {
 #'
 #' @details
 #' The \code{name} field can be one of the following:
-#' 
+#'
 #' \itemize{
 #'     \item \code{label}: label Xgboost learn from ;
 #'     \item \code{weight}: to do a weight rescale ;
 #'     \item \code{base_margin}: base margin is the base prediction Xgboost will boost from ;
 #'     \item \code{group}: number of rows in each group (to use with \code{rank:pairwise} objective).
 #' }
-#' 
+#'
 #' @examples
 #' data(agaricus.train, package='xgboost')
 #' train <- agaricus.train
 #' dtrain <- xgb.DMatrix(train$data, label=train$label)
-#' 
+#'
 #' labels <- getinfo(dtrain, 'label')
 #' setinfo(dtrain, 'label', 1-labels)
 #' labels2 <- getinfo(dtrain, 'label')
@@ -266,27 +267,27 @@ setinfo.xgb.DMatrix <- function(object, name, info, ...) {
 
 
 #' Get a new DMatrix containing the specified rows of
-#' orginal xgb.DMatrix object
+#' original xgb.DMatrix object
 #'
 #' Get a new DMatrix containing the specified rows of
-#' orginal xgb.DMatrix object
-#' 
+#' original xgb.DMatrix object
+#'
 #' @param object Object of class "xgb.DMatrix"
 #' @param idxset a integer vector of indices of rows needed
 #' @param colset currently not used (columns subsetting is not available)
 #' @param ... other parameters (currently not used)
-#' 
+#'
 #' @examples
 #' data(agaricus.train, package='xgboost')
 #' train <- agaricus.train
 #' dtrain <- xgb.DMatrix(train$data, label=train$label)
-#' 
+#'
 #' dsub <- slice(dtrain, 1:42)
 #' labels1 <- getinfo(dsub, 'label')
 #' dsub <- dtrain[1:42, ]
 #' labels2 <- getinfo(dsub, 'label')
 #' all.equal(labels1, labels2)
-#' 
+#'
 #' @rdname slice.xgb.DMatrix
 #' @export
 slice <- function(object, ...) UseMethod("slice")
@@ -301,12 +302,17 @@ slice.xgb.DMatrix <- function(object, idxset, ...) {
 
   attr_list <- attributes(object)
   nr <- nrow(object)
-  len <- sapply(attr_list, length)
+  len <- sapply(attr_list, NROW)
   ind <- which(len == nr)
   if (length(ind) > 0) {
     nms <- names(attr_list)[ind]
     for (i in seq_along(ind)) {
-      attr(ret, nms[i]) <- attr(object, nms[i])[idxset]
+      obj_attr <- attr(object, nms[i])
+      if (NCOL(obj_attr) > 1) {
+        attr(ret, nms[i]) <- obj_attr[idxset,]
+      } else {
+        attr(ret, nms[i]) <- obj_attr[idxset]
+      }
     }
   }
   return(structure(ret, class = "xgb.DMatrix"))
@@ -320,22 +326,22 @@ slice.xgb.DMatrix <- function(object, idxset, ...) {
 
 
 #' Print xgb.DMatrix
-#' 
-#' Print information about xgb.DMatrix. 
+#'
+#' Print information about xgb.DMatrix.
 #' Currently it displays dimensions and presence of info-fields and colnames.
-#' 
+#'
 #' @param x an xgb.DMatrix object
 #' @param verbose whether to print colnames (when present)
 #' @param ... not currently used
-#' 
+#'
 #' @examples
 #' data(agaricus.train, package='xgboost')
 #' train <- agaricus.train
 #' dtrain <- xgb.DMatrix(train$data, label=train$label)
-#' 
+#'
 #' dtrain
 #' print(dtrain, verbose=TRUE)
-#' 
+#'
 #' @method print xgb.DMatrix
 #' @export
 print.xgb.DMatrix <- function(x, verbose = FALSE, ...) {

R-package/R/xgb.DMatrix.save.R

@@ -11,6 +11,7 @@
 #' dtrain <- xgb.DMatrix(train$data, label=train$label)
 #' xgb.DMatrix.save(dtrain, 'xgb.DMatrix.data')
 #' dtrain <- xgb.DMatrix('xgb.DMatrix.data')
+#' if (file.exists('xgb.DMatrix.data')) file.remove('xgb.DMatrix.data')
 #' @export
 xgb.DMatrix.save <- function(dmatrix, fname) {
   if (typeof(fname) != "character")

R-package/R/xgb.create.features.R

@@ -52,9 +52,9 @@
 #' dtest <- xgb.DMatrix(data = agaricus.test$data, label = agaricus.test$label)
 #'
 #' param <- list(max_depth=2, eta=1, silent=1, objective='binary:logistic')
-#' nround = 4
+#' nrounds = 4
 #'
-#' bst = xgb.train(params = param, data = dtrain, nrounds = nround, nthread = 2)
+#' bst = xgb.train(params = param, data = dtrain, nrounds = nrounds, nthread = 2)
 #' 
 #' # Model accuracy without new features
 #' accuracy.before <- sum((predict(bst, agaricus.test$data) >= 0.5) == agaricus.test$label) /
@@ -68,7 +68,7 @@
 #' new.dtrain <- xgb.DMatrix(data = new.features.train, label = agaricus.train$label)
 #' new.dtest <- xgb.DMatrix(data = new.features.test, label = agaricus.test$label)
 #' watchlist <- list(train = new.dtrain)
-#' bst <- xgb.train(params = param, data = new.dtrain, nrounds = nround, nthread = 2)
+#' bst <- xgb.train(params = param, data = new.dtrain, nrounds = nrounds, nthread = 2)
 #' 
 #' # Model accuracy with new features
 #' accuracy.after <- sum((predict(bst, new.dtest) >= 0.5) == agaricus.test$label) /

R-package/R/xgb.cv.R

@@ -1,12 +1,12 @@
 #' Cross Validation
-#' 
+#'
 #' The cross validation function of xgboost
-#' 
+#'
 #' @param params the list of parameters. Commonly used ones are:
 #' \itemize{
 #'   \item \code{objective} objective function, common ones are
 #'   \itemize{
-#'     \item \code{reg:linear} linear regression
+#'     \item \code{reg:squarederror} Regression with squared loss
 #'     \item \code{binary:logistic} logistic regression for classification
 #'   }
 #'   \item \code{eta} step size of each boosting step
@@ -18,12 +18,12 @@
 #'   See also demo/ for walkthrough example in R.
 #' @param data takes an \code{xgb.DMatrix}, \code{matrix}, or \code{dgCMatrix} as the input.
 #' @param nrounds the max number of iterations
-#' @param nfold the original dataset is randomly partitioned into \code{nfold} equal size subsamples. 
+#' @param nfold the original dataset is randomly partitioned into \code{nfold} equal size subsamples.
 #' @param label vector of response values. Should be provided only when data is an R-matrix.
-#' @param missing is only used when input is a dense matrix. By default is set to NA, which means 
-#'        that NA values should be considered as 'missing' by the algorithm. 
+#' @param missing is only used when input is a dense matrix. By default is set to NA, which means
+#'        that NA values should be considered as 'missing' by the algorithm.
 #'        Sometimes, 0 or other extreme value might be used to represent missing values.
-#' @param prediction A logical value indicating whether to return the test fold predictions 
+#' @param prediction A logical value indicating whether to return the test fold predictions
 #'        from each CV model. This parameter engages the \code{\link{cb.cv.predict}} callback.
 #' @param showsd \code{boolean}, whether to show standard deviation of cross validation
 #' @param metrics, list of evaluation metrics to be used in cross validation,
@@ -37,22 +37,22 @@
 #'   \item \code{aucpr} Area under PR curve
 #'   \item \code{merror} Exact matching error, used to evaluate multi-class classification
 #' }
-#' @param obj customized objective function. Returns gradient and second order 
+#' @param obj customized objective function. Returns gradient and second order
 #'        gradient with given prediction and dtrain.
-#' @param feval custimized evaluation function. Returns 
-#'        \code{list(metric='metric-name', value='metric-value')} with given 
+#' @param feval customized evaluation function. Returns
+#'        \code{list(metric='metric-name', value='metric-value')} with given
 #'        prediction and dtrain.
-#' @param stratified a \code{boolean} indicating whether sampling of folds should be stratified 
+#' @param stratified a \code{boolean} indicating whether sampling of folds should be stratified
 #'        by the values of outcome labels.
 #' @param folds \code{list} provides a possibility to use a list of pre-defined CV folds
-#'        (each element must be a vector of test fold's indices). When folds are supplied, 
+#'        (each element must be a vector of test fold's indices). When folds are supplied,
 #'        the \code{nfold} and \code{stratified} parameters are ignored.
 #' @param verbose \code{boolean}, print the statistics during the process
 #' @param print_every_n Print each n-th iteration evaluation messages when \code{verbose>0}.
-#'        Default is 1 which means all messages are printed. This parameter is passed to the 
+#'        Default is 1 which means all messages are printed. This parameter is passed to the
 #'        \code{\link{cb.print.evaluation}} callback.
-#' @param early_stopping_rounds If \code{NULL}, the early stopping function is not triggered. 
-#'        If set to an integer \code{k}, training with a validation set will stop if the performance 
+#' @param early_stopping_rounds If \code{NULL}, the early stopping function is not triggered.
+#'        If set to an integer \code{k}, training with a validation set will stop if the performance
 #'        doesn't improve for \code{k} rounds.
 #'        Setting this parameter engages the \code{\link{cb.early.stop}} callback.
 #' @param maximize If \code{feval} and \code{early_stopping_rounds} are set,
@@ -60,46 +60,46 @@
 #'        When it is \code{TRUE}, it means the larger the evaluation score the better.
 #'        This parameter is passed to the \code{\link{cb.early.stop}} callback.
 #' @param callbacks a list of callback functions to perform various task during boosting.
-#'        See \code{\link{callbacks}}. Some of the callbacks are automatically created depending on the 
-#'        parameters' values. User can provide either existing or their own callback methods in order 
+#'        See \code{\link{callbacks}}. Some of the callbacks are automatically created depending on the
+#'        parameters' values. User can provide either existing or their own callback methods in order
 #'        to customize the training process.
 #' @param ... other parameters to pass to \code{params}.
-#' 
-#' @details 
-#' The original sample is randomly partitioned into \code{nfold} equal size subsamples. 
-#' 
-#' Of the \code{nfold} subsamples, a single subsample is retained as the validation data for testing the model, and the remaining \code{nfold - 1} subsamples are used as training data. 
-#' 
+#'
+#' @details
+#' The original sample is randomly partitioned into \code{nfold} equal size subsamples.
+#'
+#' Of the \code{nfold} subsamples, a single subsample is retained as the validation data for testing the model, and the remaining \code{nfold - 1} subsamples are used as training data.
+#'
 #' The cross-validation process is then repeated \code{nrounds} times, with each of the \code{nfold} subsamples used exactly once as the validation data.
-#' 
+#'
 #' All observations are used for both training and validation.
-#' 
+#'
 #' Adapted from \url{http://en.wikipedia.org/wiki/Cross-validation_\%28statistics\%29#k-fold_cross-validation}
 #'
-#' @return 
+#' @return
 #' An object of class \code{xgb.cv.synchronous} with the following elements:
 #' \itemize{
 #'   \item \code{call} a function call.
-#'   \item \code{params} parameters that were passed to the xgboost library. Note that it does not 
+#'   \item \code{params} parameters that were passed to the xgboost library. Note that it does not
 #'         capture parameters changed by the \code{\link{cb.reset.parameters}} callback.
-#'   \item \code{callbacks} callback functions that were either automatically assigned or 
+#'   \item \code{callbacks} callback functions that were either automatically assigned or
 #'         explicitly passed.
-#'   \item \code{evaluation_log} evaluation history storead as a \code{data.table} with the
-#'         first column corresponding to iteration number and the rest corresponding to the 
+#'   \item \code{evaluation_log} evaluation history stored as a \code{data.table} with the
+#'         first column corresponding to iteration number and the rest corresponding to the
 #'         CV-based evaluation means and standard deviations for the training and test CV-sets.
 #'         It is created by the \code{\link{cb.evaluation.log}} callback.
 #'   \item \code{niter} number of boosting iterations.
 #'   \item \code{nfeatures} number of features in training data.
-#'   \item \code{folds} the list of CV folds' indices - either those passed through the \code{folds} 
+#'   \item \code{folds} the list of CV folds' indices - either those passed through the \code{folds}
 #'         parameter or randomly generated.
 #'   \item \code{best_iteration} iteration number with the best evaluation metric value
 #'         (only available with early stopping).
-#'   \item \code{best_ntreelimit} the \code{ntreelimit} value corresponding to the best iteration, 
+#'   \item \code{best_ntreelimit} the \code{ntreelimit} value corresponding to the best iteration,
 #'         which could further be used in \code{predict} method
 #'         (only available with early stopping).
-#'   \item \code{pred} CV prediction values available when \code{prediction} is set. 
+#'   \item \code{pred} CV prediction values available when \code{prediction} is set.
 #'         It is either vector or matrix (see \code{\link{cb.cv.predict}}).
-#'   \item \code{models} a liost of the CV folds' models. It is only available with the explicit 
+#'   \item \code{models} a liost of the CV folds' models. It is only available with the explicit
 #'         setting of the \code{cb.cv.predict(save_models = TRUE)} callback.
 #' }
 #'
@@ -110,32 +110,32 @@
 #'                   max_depth = 3, eta = 1, objective = "binary:logistic")
 #' print(cv)
 #' print(cv, verbose=TRUE)
-#' 
+#'
 #' @export
 xgb.cv <- function(params=list(), data, nrounds, nfold, label = NULL, missing = NA,
                    prediction = FALSE, showsd = TRUE, metrics=list(),
-                   obj = NULL, feval = NULL, stratified = TRUE, folds = NULL, 
+                   obj = NULL, feval = NULL, stratified = TRUE, folds = NULL,
                    verbose = TRUE, print_every_n=1L,
                    early_stopping_rounds = NULL, maximize = NULL, callbacks = list(), ...) {
 
   check.deprecation(...)
-  
+
   params <- check.booster.params(params, ...)
   # TODO: should we deprecate the redundant 'metrics' parameter?
   for (m in metrics)
     params <- c(params, list("eval_metric" = m))
-  
+
   check.custom.obj()
   check.custom.eval()
 
   #if (is.null(params[['eval_metric']]) && is.null(feval))
   #  stop("Either 'eval_metric' or 'feval' must be provided for CV")
-  
+
   # Check the labels
   if ( (inherits(data, 'xgb.DMatrix') && is.null(getinfo(data, 'label'))) ||
        (!inherits(data, 'xgb.DMatrix') && is.null(label)))
     stop("Labels must be provided for CV either through xgb.DMatrix, or through 'label=' when 'data' is matrix")
-  
+
   # CV folds
   if(!is.null(folds)) {
     if(!is.list(folds) || length(folds) < 2)
@@ -146,7 +146,7 @@ xgb.cv <- function(params=list(), data, nrounds, nfold, label = NULL, missing =
       stop("'nfold' must be > 1")
     folds <- generate.cv.folds(nfold, nrow(data), stratified, label, params)
   }
-  
+
   # Potential TODO: sequential CV
   #if (strategy == 'sequential')
   #  stop('Sequential CV strategy is not yet implemented')
@@ -166,7 +166,7 @@ xgb.cv <- function(params=list(), data, nrounds, nfold, label = NULL, missing =
   stop_condition <- FALSE
   if (!is.null(early_stopping_rounds) &&
       !has.callbacks(callbacks, 'cb.early.stop')) {
-    callbacks <- add.cb(callbacks, cb.early.stop(early_stopping_rounds, 
+    callbacks <- add.cb(callbacks, cb.early.stop(early_stopping_rounds,
                                                  maximize = maximize, verbose = verbose))
   }
   # CV-predictions callback
@@ -177,7 +177,7 @@ xgb.cv <- function(params=list(), data, nrounds, nfold, label = NULL, missing =
   # Sort the callbacks into categories
   cb <- categorize.callbacks(callbacks)
 
-  
+
   # create the booster-folds
   dall <- xgb.get.DMatrix(data, label, missing)
   bst_folds <- lapply(seq_along(folds), function(k) {
@@ -197,12 +197,12 @@ xgb.cv <- function(params=list(), data, nrounds, nfold, label = NULL, missing =
   # those are fixed for CV (no training continuation)
   begin_iteration <- 1
   end_iteration <- nrounds
-  
+
   # synchronous CV boosting: run CV folds' models within each iteration
   for (iteration in begin_iteration:end_iteration) {
-    
+
     for (f in cb$pre_iter) f()
-    
+
     msg <- lapply(bst_folds, function(fd) {
       xgb.iter.update(fd$bst, fd$dtrain, iteration - 1, obj)
       xgb.iter.eval(fd$bst, fd$watchlist, iteration - 1, feval)
@@ -210,9 +210,9 @@ xgb.cv <- function(params=list(), data, nrounds, nfold, label = NULL, missing =
     msg <- simplify2array(msg)
     bst_evaluation <- rowMeans(msg)
     bst_evaluation_err <- sqrt(rowMeans(msg^2) - bst_evaluation^2)
-    
+
     for (f in cb$post_iter) f()
-    
+
     if (stop_condition) break
   }
   for (f in cb$finalize) f(finalize = TRUE)
@@ -236,17 +236,17 @@ xgb.cv <- function(params=list(), data, nrounds, nfold, label = NULL, missing =
 
 
 #' Print xgb.cv result
-#' 
+#'
 #' Prints formatted results of \code{xgb.cv}.
-#' 
+#'
 #' @param x an \code{xgb.cv.synchronous} object
 #' @param verbose whether to print detailed data
 #' @param ... passed to \code{data.table.print}
-#' 
+#'
 #' @details
-#' When not verbose, it would only print the evaluation results, 
+#' When not verbose, it would only print the evaluation results,
 #' including the best iteration (when available).
-#' 
+#'
 #' @examples
 #' data(agaricus.train, package='xgboost')
 #' train <- agaricus.train
@@ -254,13 +254,13 @@ xgb.cv <- function(params=list(), data, nrounds, nfold, label = NULL, missing =
 #'                eta = 1, nthread = 2, nrounds = 2, objective = "binary:logistic")
 #' print(cv)
 #' print(cv, verbose=TRUE)
-#' 
+#'
 #' @rdname print.xgb.cv
 #' @method print xgb.cv.synchronous
 #' @export
 print.xgb.cv.synchronous <- function(x, verbose = FALSE, ...) {
   cat('##### xgb.cv ', length(x$folds), '-folds\n', sep = '')
-  
+
   if (verbose) {
     if (!is.null(x$call)) {
       cat('call:\n  ')
@@ -268,8 +268,8 @@ print.xgb.cv.synchronous <- function(x, verbose = FALSE, ...) {
     }
     if (!is.null(x$params)) {
       cat('params (as set within xgb.cv):\n')
-      cat( '  ', 
-           paste(names(x$params), 
+      cat( '  ',
+           paste(names(x$params),
                  paste0('"', unlist(x$params), '"'),
                  sep = ' = ', collapse = ', '), '\n', sep = '')
     }
@@ -280,9 +280,9 @@ print.xgb.cv.synchronous <- function(x, verbose = FALSE, ...) {
         print(x)
       })
     }
-    
+
     for (n in c('niter', 'best_iteration', 'best_ntreelimit')) {
-      if (is.null(x[[n]])) 
+      if (is.null(x[[n]]))
         next
       cat(n, ': ', x[[n]], '\n', sep = '')
     }
@@ -293,10 +293,10 @@ print.xgb.cv.synchronous <- function(x, verbose = FALSE, ...) {
     }
   }
 
-  if (verbose) 
+  if (verbose)
     cat('evaluation_log:\n')
   print(x$evaluation_log, row.names = FALSE, ...)
-  
+
   if (!is.null(x$best_iteration)) {
     cat('Best iteration:\n')
     print(x$evaluation_log[x$best_iteration], row.names = FALSE, ...)

R-package/R/xgb.ggplot.R

@@ -22,7 +22,7 @@ xgb.ggplot.importance <- function(importance_matrix = NULL, top_n = NULL, measur
 
   plot <-
     ggplot2::ggplot(importance_matrix, 
-                    ggplot2::aes(x = factor(Feature, levels = rev(Feature)), y = Importance, width = 0.05),
+                    ggplot2::aes(x = factor(Feature, levels = rev(Feature)), y = Importance, width = 0.5),
                     environment = environment()) + 
     ggplot2::geom_bar(ggplot2::aes(fill = Cluster), stat = "identity", position = "identity") + 
     ggplot2::coord_flip() + 

R-package/R/xgb.load.R

@@ -28,6 +28,7 @@
 #'                eta = 1, nthread = 2, nrounds = 2,objective = "binary:logistic")
 #' xgb.save(bst, 'xgb.model')
 #' bst <- xgb.load('xgb.model')
+#' if (file.exists('xgb.model')) file.remove('xgb.model')
 #' pred <- predict(bst, test$data)
 #' @export
 xgb.load <- function(modelfile) {

R-package/R/xgb.plot.deepness.R

@@ -27,7 +27,7 @@
 #' a tree's median absolute leaf weight changes through the iterations.
 #'
 #' This function was inspired by the blog post
-#' \url{http://aysent.github.io/2015/11/08/random-forest-leaf-visualization.html}.
+#' \url{https://github.com/aysent/random-forest-leaf-visualization}.
 #' 
 #' @return
 #' 

R-package/R/xgb.plot.importance.R

@@ -5,16 +5,16 @@
 #'
 #' @param importance_matrix a \code{data.table} returned by \code{\link{xgb.importance}}.
 #' @param top_n maximal number of top features to include into the plot.
-#' @param measure the name of importance measure to plot. 
+#' @param measure the name of importance measure to plot.
 #'        When \code{NULL}, 'Gain' would be used for trees and 'Weight' would be used for gblinear.
 #' @param rel_to_first whether importance values should be represented as relative to the highest ranked feature.
 #'        See Details.
 #' @param left_margin (base R barplot) allows to adjust the left margin size to fit feature names.
 #'        When it is NULL, the existing \code{par('mar')} is used.
 #' @param cex (base R barplot) passed as \code{cex.names} parameter to \code{barplot}.
-#' @param plot (base R barplot) whether a barplot should be produced. 
+#' @param plot (base R barplot) whether a barplot should be produced.
 #'        If FALSE, only a data.table is returned.
-#' @param n_clusters (ggplot only) a \code{numeric} vector containing the min and the max range 
+#' @param n_clusters (ggplot only) a \code{numeric} vector containing the min and the max range
 #'        of the possible number of clusters of bars.
 #' @param ... other parameters passed to \code{barplot} (except horiz, border, cex.names, names.arg, and las).
 #'
@@ -22,27 +22,27 @@
 #' The graph represents each feature as a horizontal bar of length proportional to the importance of a feature.
 #' Features are shown ranked in a decreasing importance order.
 #' It works for importances from both \code{gblinear} and \code{gbtree} models.
-#' 
+#'
 #' When \code{rel_to_first = FALSE}, the values would be plotted as they were in \code{importance_matrix}.
-#' For gbtree model, that would mean being normalized to the total of 1 
+#' For gbtree model, that would mean being normalized to the total of 1
 #' ("what is feature's importance contribution relative to the whole model?").
 #' For linear models, \code{rel_to_first = FALSE} would show actual values of the coefficients.
-#' Setting \code{rel_to_first = TRUE} allows to see the picture from the perspective of 
+#' Setting \code{rel_to_first = TRUE} allows to see the picture from the perspective of
 #' "what is feature's importance contribution relative to the most important feature?"
-#' 
-#' The ggplot-backend method also performs 1-D custering of the importance values, 
-#' with bar colors coresponding to different clusters that have somewhat similar importance values.
-#' 
+#'
+#' The ggplot-backend method also performs 1-D clustering of the importance values,
+#' with bar colors corresponding to different clusters that have somewhat similar importance values.
+#'
 #' @return
 #' The \code{xgb.plot.importance} function creates a \code{barplot} (when \code{plot=TRUE})
 #' and silently returns a processed data.table with \code{n_top} features sorted by importance.
-#' 
+#'
 #' The \code{xgb.ggplot.importance} function returns a ggplot graph which could be customized afterwards.
 #' E.g., to change the title of the graph, add \code{+ ggtitle("A GRAPH NAME")} to the result.
 #'
 #' @seealso
 #' \code{\link[graphics]{barplot}}.
-#' 
+#'
 #' @examples
 #' data(agaricus.train)
 #'
@@ -50,15 +50,15 @@
 #'                eta = 1, nthread = 2, nrounds = 2, objective = "binary:logistic")
 #'
 #' importance_matrix <- xgb.importance(colnames(agaricus.train$data), model = bst)
-#' 
+#'
 #' xgb.plot.importance(importance_matrix, rel_to_first = TRUE, xlab = "Relative importance")
-#' 
+#'
 #' (gg <- xgb.ggplot.importance(importance_matrix, measure = "Frequency", rel_to_first = TRUE))
 #' gg + ggplot2::ylab("Frequency")
 #'
 #' @rdname xgb.plot.importance
 #' @export
-xgb.plot.importance <- function(importance_matrix = NULL, top_n = NULL, measure = NULL, 
+xgb.plot.importance <- function(importance_matrix = NULL, top_n = NULL, measure = NULL,
                                 rel_to_first = FALSE, left_margin = 10, cex = NULL, plot = TRUE, ...) {
   check.deprecation(...)
   if (!is.data.table(importance_matrix))  {
@@ -80,13 +80,13 @@ xgb.plot.importance <- function(importance_matrix = NULL, top_n = NULL, measure
     if (!"Feature" %in% imp_names)
       stop("Importance matrix column names are not as expected!")
   }
-  
+
   # also aggregate, just in case when the values were not yet summed up by feature
   importance_matrix <- importance_matrix[, Importance := sum(get(measure)), by = Feature]
-  
+
   # make sure it's ordered
   importance_matrix <- importance_matrix[order(-abs(Importance))]
-  
+
   if (!is.null(top_n)) {
     top_n <- min(top_n, nrow(importance_matrix))
     importance_matrix <- head(importance_matrix, top_n)
@@ -97,14 +97,14 @@ xgb.plot.importance <- function(importance_matrix = NULL, top_n = NULL, measure
   if (is.null(cex)) {
     cex <- 2.5/log2(1 + nrow(importance_matrix))
   }
-  
+
   if (plot) {
     op <- par(no.readonly = TRUE)
     mar <- op$mar
     if (!is.null(left_margin))
       mar[2] <- left_margin
     par(mar = mar)
-    
+
     # reverse the order of rows to have the highest ranked at the top
     importance_matrix[nrow(importance_matrix):1,
                       barplot(Importance, horiz = TRUE, border = NA, cex.names = cex,
@@ -115,7 +115,7 @@ xgb.plot.importance <- function(importance_matrix = NULL, top_n = NULL, measure
                       barplot(Importance, horiz = TRUE, border = NA, add = TRUE)]
     par(op)
   }
-  
+
   invisible(importance_matrix)
 }
 

R-package/R/xgb.plot.shap.R

@@ -1,9 +1,9 @@
 #' SHAP contribution dependency plots
 #'
 #' Visualizing the SHAP feature contribution to prediction dependencies on feature value.
-#' 
+#'
 #' @param data data as a \code{matrix} or \code{dgCMatrix}.
-#' @param shap_contrib a matrix of SHAP contributions that was computed earlier for the above 
+#' @param shap_contrib a matrix of SHAP contributions that was computed earlier for the above
 #'          \code{data}. When it is NULL, it is computed internally using \code{model} and \code{data}.
 #' @param features a vector of either column indices or of feature names to plot. When it is NULL,
 #'          feature importance is calculated, and \code{top_n} high ranked features are taken.
@@ -31,32 +31,32 @@
 #' @param plot_loess whether to plot loess-smoothed curves. The smoothing is only done for features with
 #'          more than 5 distinct values.
 #' @param col_loess a color to use for the loess curves.
-#' @param span_loess the \code{span} paramerer in \code{\link[stats]{loess}}'s call.
+#' @param span_loess the \code{span} parameter in \code{\link[stats]{loess}}'s call.
 #' @param which whether to do univariate or bivariate plotting. NOTE: only 1D is implemented so far.
 #' @param plot whether a plot should be drawn. If FALSE, only a lits of matrices is returned.
 #' @param ... other parameters passed to \code{plot}.
-#' 
+#'
 #' @details
-#' 
+#'
 #' These scatterplots represent how SHAP feature contributions depend of feature values.
 #' The similarity to partial dependency plots is that they also give an idea for how feature values
 #' affect predictions. However, in partial dependency plots, we usually see marginal dependencies
 #' of model prediction on feature value, while SHAP contribution dependency plots display the estimated
 #' contributions of a feature to model prediction for each individual case.
-#' 
+#'
 #' When \code{plot_loess = TRUE} is set, feature values are rounded to 3 significant digits and
 #' weighted LOESS is computed and plotted, where weights are the numbers of data points
 #' at each rounded value.
-#' 
+#'
 #' Note: SHAP contributions are shown on the scale of model margin. E.g., for a logistic binomial objective,
 #' the margin is prediction before a sigmoidal transform into probability-like values.
 #' Also, since SHAP stands for "SHapley Additive exPlanation" (model prediction = sum of SHAP
 #' contributions for all features + bias), depending on the objective used, transforming SHAP
 #' contributions for a feature from the marginal to the prediction space is not necessarily
 #' a meaningful thing to do.
-#' 
+#'
 #' @return
-#' 
+#'
 #' In addition to producing plots (when \code{plot=TRUE}), it silently returns a list of two matrices:
 #' \itemize{
 #'  \item \code{data} the values of selected features;
@@ -70,11 +70,11 @@
 #' Scott M. Lundberg, Su-In Lee, "Consistent feature attribution for tree ensembles", \url{https://arxiv.org/abs/1706.06060}
 #'
 #' @examples
-#' 
+#'
 #' data(agaricus.train, package='xgboost')
 #' data(agaricus.test, package='xgboost')
 #'
-#' bst <- xgboost(agaricus.train$data, agaricus.train$label, nrounds = 50, 
+#' bst <- xgboost(agaricus.train$data, agaricus.train$label, nrounds = 50,
 #'                eta = 0.1, max_depth = 3, subsample = .5,
 #'                method = "hist", objective = "binary:logistic", nthread = 2, verbose = 0)
 #'
@@ -99,7 +99,7 @@
 #'               n_col = 2, col = col, pch = 16, pch_NA = 17)
 #' xgb.plot.shap(x, model = mbst, trees = trees0 + 2, target_class = 2, top_n = 4,
 #'               n_col = 2, col = col, pch = 16, pch_NA = 17)
-#' 
+#'
 #' @rdname xgb.plot.shap
 #' @export
 xgb.plot.shap <- function(data, shap_contrib = NULL, features = NULL, top_n = 1, model = NULL,
@@ -109,7 +109,7 @@ xgb.plot.shap <- function(data, shap_contrib = NULL, features = NULL, top_n = 1,
                           plot_NA = TRUE, col_NA = rgb(0.7, 0, 1, 0.6), pch_NA = '.', pos_NA = 1.07,
                           plot_loess = TRUE, col_loess = 2, span_loess = 0.5,
                           which = c("1d", "2d"), plot = TRUE, ...) {
-  
+
   if (!is.matrix(data) && !inherits(data, "dgCMatrix"))
     stop("data: must be either matrix or dgCMatrix")
 
@@ -122,7 +122,7 @@ xgb.plot.shap <- function(data, shap_contrib = NULL, features = NULL, top_n = 1,
   if (!is.null(shap_contrib) &&
       (!is.matrix(shap_contrib) || nrow(shap_contrib) != nrow(data) || ncol(shap_contrib) != ncol(data) + 1))
     stop("shap_contrib is not compatible with the provided data")
-  
+
   nsample <- if (is.null(subsample)) min(100000, nrow(data)) else as.integer(subsample * nrow(data))
   idx <- sample(1:nrow(data), nsample)
   data <- data[idx,]
@@ -144,13 +144,13 @@ xgb.plot.shap <- function(data, shap_contrib = NULL, features = NULL, top_n = 1,
       stop("top_n: must be an integer within [1, 100]")
     features <- imp$Feature[1:min(top_n, NROW(imp))]
   }
-  
+
   if (is.character(features)) {
     if (is.null(colnames(data)))
       stop("Either provide `data` with column names or provide `features` as column indices")
     features <- match(features, colnames(data))
   }
-  
+
   if (n_col > length(features)) n_col <- length(features)
 
   if (is.list(shap_contrib)) { # multiclass: either choose a class or merge
@@ -165,7 +165,7 @@ xgb.plot.shap <- function(data, shap_contrib = NULL, features = NULL, top_n = 1,
   if (is.null(cols)) cols <- paste0('X', 1:ncol(data))
   colnames(data) <- cols
   colnames(shap_contrib) <- cols
-  
+
   if (plot && which == "1d") {
     op <- par(mfrow = c(ceiling(length(features) / n_col), n_col),
               oma = c(0,0,0,0) + 0.2,

R-package/R/xgb.save.R

@@ -27,6 +27,7 @@
 #'                eta = 1, nthread = 2, nrounds = 2,objective = "binary:logistic")
 #' xgb.save(bst, 'xgb.model')
 #' bst <- xgb.load('xgb.model')
+#' if (file.exists('xgb.model')) file.remove('xgb.model')
 #' pred <- predict(bst, test$data)
 #' @export
 xgb.save <- function(model, fname) {

R-package/R/xgb.train.R

@@ -1,47 +1,48 @@
 #' eXtreme Gradient Boosting Training
-#' 
+#'
 #' \code{xgb.train} is an advanced interface for training an xgboost model.
 #' The \code{xgboost} function is a simpler wrapper for \code{xgb.train}.
 #'
-#' @param params the list of parameters. 
+#' @param params the list of parameters.
 #'        The complete list of parameters is available at \url{http://xgboost.readthedocs.io/en/latest/parameter.html}.
 #'        Below is a shorter summary:
-#' 
+#'
 #' 1. General Parameters
-#' 
+#'
 #' \itemize{
 #'   \item \code{booster} which booster to use, can be \code{gbtree} or \code{gblinear}. Default: \code{gbtree}.
 #' }
-#'  
+#'
 #' 2. Booster Parameters
-#' 
+#'
 #' 2.1. Parameter for Tree Booster
-#' 
+#'
 #' \itemize{
 #'   \item \code{eta} control the learning rate: scale the contribution of each tree by a factor of \code{0 < eta < 1} when it is added to the current approximation. Used to prevent overfitting by making the boosting process more conservative. Lower value for \code{eta} implies larger value for \code{nrounds}: low \code{eta} value means model more robust to overfitting but slower to compute. Default: 0.3
-#'   \item \code{gamma} minimum loss reduction required to make a further partition on a leaf node of the tree. the larger, the more conservative the algorithm will be. 
+#'   \item \code{gamma} minimum loss reduction required to make a further partition on a leaf node of the tree. the larger, the more conservative the algorithm will be.
 #'   \item \code{max_depth} maximum depth of a tree. Default: 6
 #'   \item \code{min_child_weight} minimum sum of instance weight (hessian) needed in a child. If the tree partition step results in a leaf node with the sum of instance weight less than min_child_weight, then the building process will give up further partitioning. In linear regression mode, this simply corresponds to minimum number of instances needed to be in each node. The larger, the more conservative the algorithm will be. Default: 1
-#'   \item \code{subsample} subsample ratio of the training instance. Setting it to 0.5 means that xgboost randomly collected half of the data instances to grow trees and this will prevent overfitting. It makes computation shorter (because less data to analyse). It is advised to use this parameter with \code{eta} and increase \code{nround}. Default: 1 
+#'   \item \code{subsample} subsample ratio of the training instance. Setting it to 0.5 means that xgboost randomly collected half of the data instances to grow trees and this will prevent overfitting. It makes computation shorter (because less data to analyse). It is advised to use this parameter with \code{eta} and increase \code{nrounds}. Default: 1
 #'   \item \code{colsample_bytree} subsample ratio of columns when constructing each tree. Default: 1
 #'   \item \code{num_parallel_tree} Experimental parameter. number of trees to grow per round. Useful to test Random Forest through Xgboost (set \code{colsample_bytree < 1}, \code{subsample  < 1}  and \code{round = 1}) accordingly. Default: 1
 #'   \item \code{monotone_constraints} A numerical vector consists of \code{1}, \code{0} and \code{-1} with its length equals to the number of features in the training data. \code{1} is increasing, \code{-1} is decreasing and \code{0} is no constraint.
+#'   \item \code{interaction_constraints} A list of vectors specifying feature indices of permitted interactions. Each item of the list represents one permitted interaction where specified features are allowed to interact with each other. Feature index values should start from \code{0} (\code{0} references the first column).  Leave argument unspecified for no interaction constraints.
 #' }
-#' 
+#'
 #' 2.2. Parameter for Linear Booster
-#'  
+#'
 #' \itemize{
 #'   \item \code{lambda} L2 regularization term on weights. Default: 0
 #'   \item \code{lambda_bias} L2 regularization term on bias. Default: 0
 #'   \item \code{alpha} L1 regularization term on weights. (there is no L1 reg on bias because it is not important). Default: 0
 #' }
-#' 
-#' 3. Task Parameters 
-#' 
+#'
+#' 3. Task Parameters
+#'
 #' \itemize{
 #' \item \code{objective} specify the learning task and the corresponding learning objective, users can pass a self-defined function to it. The default objective options are below:
 #'   \itemize{
-#'     \item \code{reg:linear} linear regression (Default).
+#'     \item \code{reg:squarederror} Regression with squared loss (Default).
 #'     \item \code{reg:logistic} logistic regression.
 #'     \item \code{binary:logistic} logistic regression for binary classification. Output probability.
 #'     \item \code{binary:logitraw} logistic regression for binary classification, output score before logistic transformation.
@@ -53,32 +54,32 @@
 #'   \item \code{base_score} the initial prediction score of all instances, global bias. Default: 0.5
 #'   \item \code{eval_metric} evaluation metrics for validation data. Users can pass a self-defined function to it. Default: metric will be assigned according to objective(rmse for regression, and error for classification, mean average precision for ranking). List is provided in detail section.
 #' }
-#' 
+#'
 #' @param data training dataset. \code{xgb.train} accepts only an \code{xgb.DMatrix} as the input.
 #'        \code{xgboost}, in addition, also accepts \code{matrix}, \code{dgCMatrix}, or name of a local data file.
 #' @param nrounds max number of boosting iterations.
 #' @param watchlist named list of xgb.DMatrix datasets to use for evaluating model performance.
 #'        Metrics specified in either \code{eval_metric} or \code{feval} will be computed for each
-#'        of these datasets during each boosting iteration, and stored in the end as a field named 
-#'        \code{evaluation_log} in the resulting object. When either \code{verbose>=1} or 
+#'        of these datasets during each boosting iteration, and stored in the end as a field named
+#'        \code{evaluation_log} in the resulting object. When either \code{verbose>=1} or
 #'        \code{\link{cb.print.evaluation}} callback is engaged, the performance results are continuously
-#'        printed out during the training. 
+#'        printed out during the training.
 #'        E.g., specifying \code{watchlist=list(validation1=mat1, validation2=mat2)} allows to track
 #'        the performance of each round's model on mat1 and mat2.
-#' @param obj customized objective function. Returns gradient and second order 
+#' @param obj customized objective function. Returns gradient and second order
 #'        gradient with given prediction and dtrain.
-#' @param feval custimized evaluation function. Returns 
-#'        \code{list(metric='metric-name', value='metric-value')} with given 
+#' @param feval customized evaluation function. Returns
+#'        \code{list(metric='metric-name', value='metric-value')} with given
 #'        prediction and dtrain.
 #' @param verbose If 0, xgboost will stay silent. If 1, it will print information about performance.
 #'        If 2, some additional information will be printed out.
-#'        Note that setting \code{verbose > 0} automatically engages the 
+#'        Note that setting \code{verbose > 0} automatically engages the
 #'        \code{cb.print.evaluation(period=1)} callback function.
 #' @param print_every_n Print each n-th iteration evaluation messages when \code{verbose>0}.
-#'        Default is 1 which means all messages are printed. This parameter is passed to the 
+#'        Default is 1 which means all messages are printed. This parameter is passed to the
 #'        \code{\link{cb.print.evaluation}} callback.
-#' @param early_stopping_rounds If \code{NULL}, the early stopping function is not triggered. 
-#'        If set to an integer \code{k}, training with a validation set will stop if the performance 
+#' @param early_stopping_rounds If \code{NULL}, the early stopping function is not triggered.
+#'        If set to an integer \code{k}, training with a validation set will stop if the performance
 #'        doesn't improve for \code{k} rounds.
 #'        Setting this parameter engages the \code{\link{cb.early.stop}} callback.
 #' @param maximize If \code{feval} and \code{early_stopping_rounds} are set,
@@ -89,35 +90,35 @@
 #'        0 means save at the end. The saving is handled by the \code{\link{cb.save.model}} callback.
 #' @param save_name the name or path for periodically saved model file.
 #' @param xgb_model a previously built model to continue the training from.
-#'        Could be either an object of class \code{xgb.Booster}, or its raw data, or the name of a 
+#'        Could be either an object of class \code{xgb.Booster}, or its raw data, or the name of a
 #'        file with a previously saved model.
 #' @param callbacks a list of callback functions to perform various task during boosting.
-#'        See \code{\link{callbacks}}. Some of the callbacks are automatically created depending on the 
-#'        parameters' values. User can provide either existing or their own callback methods in order 
+#'        See \code{\link{callbacks}}. Some of the callbacks are automatically created depending on the
+#'        parameters' values. User can provide either existing or their own callback methods in order
 #'        to customize the training process.
 #' @param ... other parameters to pass to \code{params}.
-#' @param label vector of response values. Should not be provided when data is 
+#' @param label vector of response values. Should not be provided when data is
 #'        a local data file name or an \code{xgb.DMatrix}.
 #' @param missing by default is set to NA, which means that NA values should be considered as 'missing'
 #'        by the algorithm. Sometimes, 0 or other extreme value might be used to represent missing values.
 #'        This parameter is only used when input is a dense matrix.
 #' @param weight a vector indicating the weight for each row of the input.
-#' 
-#' @details 
-#' These are the training functions for \code{xgboost}. 
-#' 
-#' The \code{xgb.train} interface supports advanced features such as \code{watchlist}, 
-#' customized objective and evaluation metric functions, therefore it is more flexible 
+#'
+#' @details
+#' These are the training functions for \code{xgboost}.
+#'
+#' The \code{xgb.train} interface supports advanced features such as \code{watchlist},
+#' customized objective and evaluation metric functions, therefore it is more flexible
 #' than the \code{xgboost} interface.
 #'
-#' Parallelization is automatically enabled if \code{OpenMP} is present. 
+#' Parallelization is automatically enabled if \code{OpenMP} is present.
 #' Number of threads can also be manually specified via \code{nthread} parameter.
-#' 
+#'
 #' The evaluation metric is chosen automatically by Xgboost (according to the objective)
 #' when the \code{eval_metric} parameter is not provided.
-#' User may set one or several \code{eval_metric} parameters. 
+#' User may set one or several \code{eval_metric} parameters.
 #' Note that when using a customized metric, only this single metric can be used.
-#' The folloiwing is the list of built-in metrics for which Xgboost provides optimized implementation:
+#' The following is the list of built-in metrics for which Xgboost provides optimized implementation:
 #'   \itemize{
 #'      \item \code{rmse} root mean square error. \url{http://en.wikipedia.org/wiki/Root_mean_square_error}
 #'      \item \code{logloss} negative log-likelihood. \url{http://en.wikipedia.org/wiki/Log-likelihood}
@@ -130,7 +131,7 @@
 #'      \item \code{aucpr} Area under the PR curve. \url{https://en.wikipedia.org/wiki/Precision_and_recall} for ranking evaluation.
 #'      \item \code{ndcg} Normalized Discounted Cumulative Gain (for ranking task). \url{http://en.wikipedia.org/wiki/NDCG}
 #'   }
-#' 
+#'
 #' The following callbacks are automatically created when certain parameters are set:
 #' \itemize{
 #'   \item \code{cb.print.evaluation} is turned on when \code{verbose > 0};
@@ -139,38 +140,38 @@
 #'   \item \code{cb.early.stop}: when \code{early_stopping_rounds} is set.
 #'   \item \code{cb.save.model}: when \code{save_period > 0} is set.
 #' }
-#' 
-#' @return 
+#'
+#' @return
 #' An object of class \code{xgb.Booster} with the following elements:
 #' \itemize{
 #'   \item \code{handle} a handle (pointer) to the xgboost model in memory.
 #'   \item \code{raw} a cached memory dump of the xgboost model saved as R's \code{raw} type.
 #'   \item \code{niter} number of boosting iterations.
-#'   \item \code{evaluation_log} evaluation history storead as a \code{data.table} with the
+#'   \item \code{evaluation_log} evaluation history stored as a \code{data.table} with the
 #'         first column corresponding to iteration number and the rest corresponding to evaluation
 #'         metrics' values. It is created by the \code{\link{cb.evaluation.log}} callback.
 #'   \item \code{call} a function call.
-#'   \item \code{params} parameters that were passed to the xgboost library. Note that it does not 
+#'   \item \code{params} parameters that were passed to the xgboost library. Note that it does not
 #'         capture parameters changed by the \code{\link{cb.reset.parameters}} callback.
-#'   \item \code{callbacks} callback functions that were either automatically assigned or 
-#'         explicitely passed.
+#'   \item \code{callbacks} callback functions that were either automatically assigned or
+#'         explicitly passed.
 #'   \item \code{best_iteration} iteration number with the best evaluation metric value
 #'         (only available with early stopping).
-#'   \item \code{best_ntreelimit} the \code{ntreelimit} value corresponding to the best iteration, 
+#'   \item \code{best_ntreelimit} the \code{ntreelimit} value corresponding to the best iteration,
 #'         which could further be used in \code{predict} method
 #'         (only available with early stopping).
 #'   \item \code{best_score} the best evaluation metric value during early stopping.
 #'         (only available with early stopping).
 #'   \item \code{feature_names} names of the training dataset features
-#'         (only when comun names were defined in training data).
+#'         (only when column names were defined in training data).
 #'   \item \code{nfeatures} number of features in training data.
 #' }
-#' 
+#'
 #' @seealso
 #' \code{\link{callbacks}},
 #' \code{\link{predict.xgb.Booster}},
 #' \code{\link{xgb.cv}}
-#' 
+#'
 #' @references
 #'
 #' Tianqi Chen and Carlos Guestrin, "XGBoost: A Scalable Tree Boosting System",
@@ -179,17 +180,17 @@
 #' @examples
 #' data(agaricus.train, package='xgboost')
 #' data(agaricus.test, package='xgboost')
-#' 
+#'
 #' dtrain <- xgb.DMatrix(agaricus.train$data, label = agaricus.train$label)
 #' dtest <- xgb.DMatrix(agaricus.test$data, label = agaricus.test$label)
 #' watchlist <- list(train = dtrain, eval = dtest)
-#' 
+#'
 #' ## A simple xgb.train example:
-#' param <- list(max_depth = 2, eta = 1, silent = 1, nthread = 2, 
+#' param <- list(max_depth = 2, eta = 1, verbose = 0, nthread = 2,
 #'               objective = "binary:logistic", eval_metric = "auc")
 #' bst <- xgb.train(param, dtrain, nrounds = 2, watchlist)
-#' 
-#' 
+#'
+#'
 #' ## An xgb.train example where custom objective and evaluation metric are used:
 #' logregobj <- function(preds, dtrain) {
 #'    labels <- getinfo(dtrain, "label")
@@ -203,58 +204,58 @@
 #'   err <- as.numeric(sum(labels != (preds > 0)))/length(labels)
 #'   return(list(metric = "error", value = err))
 #' }
-#' 
+#'
 #' # These functions could be used by passing them either:
 #' #  as 'objective' and 'eval_metric' parameters in the params list:
-#' param <- list(max_depth = 2, eta = 1, silent = 1, nthread = 2, 
+#' param <- list(max_depth = 2, eta = 1, verbose = 0, nthread = 2,
 #'               objective = logregobj, eval_metric = evalerror)
 #' bst <- xgb.train(param, dtrain, nrounds = 2, watchlist)
-#' 
+#'
 #' #  or through the ... arguments:
-#' param <- list(max_depth = 2, eta = 1, silent = 1, nthread = 2)
+#' param <- list(max_depth = 2, eta = 1, verbose = 0, nthread = 2)
 #' bst <- xgb.train(param, dtrain, nrounds = 2, watchlist,
 #'                  objective = logregobj, eval_metric = evalerror)
-#' 
+#'
 #' #  or as dedicated 'obj' and 'feval' parameters of xgb.train:
 #' bst <- xgb.train(param, dtrain, nrounds = 2, watchlist,
 #'                  obj = logregobj, feval = evalerror)
-#' 
-#' 
+#'
+#'
 #' ## An xgb.train example of using variable learning rates at each iteration:
-#' param <- list(max_depth = 2, eta = 1, silent = 1, nthread = 2,
+#' param <- list(max_depth = 2, eta = 1, verbose = 0, nthread = 2,
 #'               objective = "binary:logistic", eval_metric = "auc")
 #' my_etas <- list(eta = c(0.5, 0.1))
 #' bst <- xgb.train(param, dtrain, nrounds = 2, watchlist,
 #'                  callbacks = list(cb.reset.parameters(my_etas)))
-#' 
+#'
 #' ## Early stopping:
 #' bst <- xgb.train(param, dtrain, nrounds = 25, watchlist,
 #'                  early_stopping_rounds = 3)
-#' 
+#'
 #' ## An 'xgboost' interface example:
-#' bst <- xgboost(data = agaricus.train$data, label = agaricus.train$label, 
-#'                max_depth = 2, eta = 1, nthread = 2, nrounds = 2, 
+#' bst <- xgboost(data = agaricus.train$data, label = agaricus.train$label,
+#'                max_depth = 2, eta = 1, nthread = 2, nrounds = 2,
 #'                objective = "binary:logistic")
 #' pred <- predict(bst, agaricus.test$data)
-#' 
+#'
 #' @rdname xgb.train
 #' @export
 xgb.train <- function(params = list(), data, nrounds, watchlist = list(),
                       obj = NULL, feval = NULL, verbose = 1, print_every_n = 1L,
                       early_stopping_rounds = NULL, maximize = NULL,
-                      save_period = NULL, save_name = "xgboost.model", 
+                      save_period = NULL, save_name = "xgboost.model",
                       xgb_model = NULL, callbacks = list(), ...) {
-  
+
   check.deprecation(...)
-  
+
   params <- check.booster.params(params, ...)
 
   check.custom.obj()
   check.custom.eval()
-  
+
   # data & watchlist checks
   dtrain <- data
-  if (!inherits(dtrain, "xgb.DMatrix")) 
+  if (!inherits(dtrain, "xgb.DMatrix"))
     stop("second argument dtrain must be xgb.DMatrix")
   if (length(watchlist) > 0) {
     if (typeof(watchlist) != "list" ||
@@ -287,7 +288,7 @@ xgb.train <- function(params = list(), data, nrounds, watchlist = list(),
   stop_condition <- FALSE
   if (!is.null(early_stopping_rounds) &&
       !has.callbacks(callbacks, 'cb.early.stop')) {
-    callbacks <- add.cb(callbacks, cb.early.stop(early_stopping_rounds, 
+    callbacks <- add.cb(callbacks, cb.early.stop(early_stopping_rounds,
                                                  maximize = maximize, verbose = verbose))
   }
   # Sort the callbacks into categories
@@ -317,22 +318,22 @@ xgb.train <- function(params = list(), data, nrounds, watchlist = list(),
 
   # TODO: distributed code
   rank <- 0
-  
+
   niter_skip <- ifelse(is_update, 0, niter_init)
   begin_iteration <- niter_skip + 1
   end_iteration <- niter_skip + nrounds
-  
+
   # the main loop for boosting iterations
   for (iteration in begin_iteration:end_iteration) {
-    
+
     for (f in cb$pre_iter) f()
-    
+
     xgb.iter.update(bst$handle, dtrain, iteration - 1, obj)
-    
+
     bst_evaluation <- numeric(0)
     if (length(watchlist) > 0)
       bst_evaluation <- xgb.iter.eval(bst$handle, watchlist, iteration - 1, feval)
-    
+
     xgb.attr(bst$handle, 'niter') <- iteration - 1
 
     for (f in cb$post_iter) f()
@@ -340,9 +341,9 @@ xgb.train <- function(params = list(), data, nrounds, watchlist = list(),
     if (stop_condition) break
   }
   for (f in cb$finalize) f(finalize = TRUE)
-  
+
   bst <- xgb.Booster.complete(bst, saveraw = TRUE)
-  
+
   # store the total number of boosting iterations
   bst$niter = end_iteration
 

R-package/README.md

@@ -30,4 +30,4 @@ Examples
 Development
 -----------
 
-* See the [R Package section](https://xgboost.readthedocs.io/en/latest/how_to/contribute.html#r-package) of the contributors guide.
+* See the [R Package section](https://xgboost.readthedocs.io/en/latest/contribute.html#r-package) of the contributors guide.

R-package/cleanup

@@ -1,3 +1,4 @@
 #!/bin/sh
 
 rm -f src/Makevars
+rm -f CMakeLists.txt

R-package/configure

@@ -1667,12 +1667,12 @@ OPENMP_CXXFLAGS=""
 
 if test `uname -s` = "Linux"
 then
-  OPENMP_CXXFLAGS="\$(SHLIB_OPENMP_CFLAGS)"
+  OPENMP_CXXFLAGS="\$(SHLIB_OPENMP_CXXFLAGS)"
 fi
 
 if test `uname -s` = "Darwin"
 then
-  OPENMP_CXXFLAGS="\$(SHLIB_OPENMP_CFLAGS)"
+  OPENMP_CXXFLAGS="\$(SHLIB_OPENMP_CXXFLAGS)"
   ac_pkg_openmp=no
   { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether OpenMP will work in a package" >&5
 $as_echo_n "checking whether OpenMP will work in a package... " >&6; }

R-package/configure.ac

@@ -8,12 +8,12 @@ OPENMP_CXXFLAGS=""
 
 if test `uname -s` = "Linux"
 then
-  OPENMP_CXXFLAGS="\$(SHLIB_OPENMP_CFLAGS)"
+  OPENMP_CXXFLAGS="\$(SHLIB_OPENMP_CXXFLAGS)"
 fi
 
 if test `uname -s` = "Darwin"
 then
-  OPENMP_CXXFLAGS="\$(SHLIB_OPENMP_CFLAGS)"
+  OPENMP_CXXFLAGS="\$(SHLIB_OPENMP_CXXFLAGS)"
   ac_pkg_openmp=no
   AC_MSG_CHECKING([whether OpenMP will work in a package])
   AC_LANG_CONFTEST(

R-package/demo/00Index

@@ -11,4 +11,5 @@ early_stopping                  Early Stop in training
 poisson_regression              Poisson Regression on count data
 tweedie_regression              Tweddie Regression
 gpu_accelerated                 GPU-accelerated tree building algorithms
+interaction_constraints         Interaction constraints among features
 

R-package/demo/cross_validation.R

@@ -5,20 +5,20 @@ data(agaricus.test, package='xgboost')
 dtrain <- xgb.DMatrix(agaricus.train$data, label = agaricus.train$label)
 dtest <- xgb.DMatrix(agaricus.test$data, label = agaricus.test$label)
 
-nround <- 2
+nrounds <- 2
 param <- list(max_depth=2, eta=1, silent=1, nthread=2, objective='binary:logistic')
 
 cat('running cross validation\n')
 # do cross validation, this will print result out as
 # [iteration]  metric_name:mean_value+std_value
 # std_value is standard deviation of the metric
-xgb.cv(param, dtrain, nround, nfold=5, metrics={'error'})
+xgb.cv(param, dtrain, nrounds, nfold=5, metrics={'error'})
 
 cat('running cross validation, disable standard deviation display\n')
 # do cross validation, this will print result out as
 # [iteration]  metric_name:mean_value+std_value
 # std_value is standard deviation of the metric
-xgb.cv(param, dtrain, nround, nfold=5,
+xgb.cv(param, dtrain, nrounds, nfold=5,
        metrics='error', showsd = FALSE)
 
 ###
@@ -43,9 +43,9 @@ evalerror <- function(preds, dtrain) {
 param <- list(max_depth=2, eta=1, silent=1,
               objective = logregobj, eval_metric = evalerror)
 # train with customized objective
-xgb.cv(params = param, data = dtrain, nrounds = nround, nfold = 5)
+xgb.cv(params = param, data = dtrain, nrounds = nrounds, nfold = 5)
 
 # do cross validation with prediction values for each fold
-res <- xgb.cv(params = param, data = dtrain, nrounds = nround, nfold = 5, prediction = TRUE)
+res <- xgb.cv(params = param, data = dtrain, nrounds = nrounds, nfold = 5, prediction = TRUE)
 res$evaluation_log
 length(res$pred)

R-package/demo/custom_objective.R

@@ -33,7 +33,7 @@ evalerror <- function(preds, dtrain) {
   return(list(metric = "error", value = err))
 }
 
-param <- list(max_depth=2, eta=1, nthread = 2, silent=1, 
+param <- list(max_depth=2, eta=1, nthread = 2, verbosity=0, 
               objective=logregobj, eval_metric=evalerror)
 print ('start training with user customized objective')
 # training with customized objective, we can also do step by step training
@@ -57,7 +57,7 @@ logregobjattr <- function(preds, dtrain) {
   hess <- preds * (1 - preds)
   return(list(grad = grad, hess = hess))
 }
-param <- list(max_depth=2, eta=1, nthread = 2, silent=1, 
+param <- list(max_depth=2, eta=1, nthread = 2, verbosity=0, 
               objective=logregobjattr, eval_metric=evalerror)
 print ('start training with user customized objective, with additional attributes in DMatrix')
 # training with customized objective, we can also do step by step training

R-package/demo/early_stopping.R

@@ -7,7 +7,7 @@ dtest <- xgb.DMatrix(agaricus.test$data, label = agaricus.test$label)
 # note: for customized objective function, we leave objective as default
 # note: what we are getting is margin value in prediction
 # you must know what you are doing
-param <- list(max_depth=2, eta=1, nthread = 2, silent=1)
+param <- list(max_depth=2, eta=1, nthread=2, verbosity=0)
 watchlist <- list(eval = dtest)
 num_round <- 20
 # user define objective function, given prediction, return gradient and second order gradient
@@ -32,9 +32,9 @@ evalerror <- function(preds, dtrain) {
 }
 print ('start training with early Stopping setting')
 
-bst <- xgb.train(param, dtrain, num_round, watchlist, 
+bst <- xgb.train(param, dtrain, num_round, watchlist,
                  objective = logregobj, eval_metric = evalerror, maximize = FALSE,
                  early_stopping_round = 3)
-bst <- xgb.cv(param, dtrain, num_round, nfold = 5, 
+bst <- xgb.cv(param, dtrain, num_round, nfold = 5,
               objective = logregobj, eval_metric = evalerror,
               maximize = FALSE, early_stopping_rounds = 3)

R-package/demo/interaction_constraints.R

@@ -0,0 +1,105 @@
+library(xgboost)
+library(data.table)
+
+set.seed(1024)
+
+# Function to obtain a list of interactions fitted in trees, requires input of maximum depth
+treeInteractions <- function(input_tree, input_max_depth){
+  trees <- copy(input_tree)  # copy tree input to prevent overwriting
+  if (input_max_depth < 2) return(list())  # no interactions if max depth < 2
+  if (nrow(input_tree) == 1) return(list())
+
+  # Attach parent nodes
+  for (i in 2:input_max_depth){
+    if (i == 2) trees[, ID_merge:=ID] else trees[, ID_merge:=get(paste0('parent_',i-2))]
+    parents_left <- trees[!is.na(Split), list(i.id=ID, i.feature=Feature, ID_merge=Yes)]
+    parents_right <- trees[!is.na(Split), list(i.id=ID, i.feature=Feature, ID_merge=No)]
+
+    setorderv(trees, 'ID_merge')
+    setorderv(parents_left, 'ID_merge')
+    setorderv(parents_right, 'ID_merge')
+
+    trees <- merge(trees, parents_left, by='ID_merge', all.x=T)
+    trees[!is.na(i.id), c(paste0('parent_', i-1), paste0('parent_feat_', i-1)):=list(i.id, i.feature)]
+    trees[, c('i.id','i.feature'):=NULL]
+
+    trees <- merge(trees, parents_right, by='ID_merge', all.x=T)
+    trees[!is.na(i.id), c(paste0('parent_', i-1), paste0('parent_feat_', i-1)):=list(i.id, i.feature)]
+    trees[, c('i.id','i.feature'):=NULL]
+  }
+
+  # Extract nodes with interactions
+  interaction_trees <- trees[!is.na(Split) & !is.na(parent_1), 
+                             c('Feature',paste0('parent_feat_',1:(input_max_depth-1))), with=F]
+  interaction_trees_split <- split(interaction_trees, 1:nrow(interaction_trees))
+  interaction_list <- lapply(interaction_trees_split, as.character)
+
+  # Remove NAs (no parent interaction)
+  interaction_list <- lapply(interaction_list, function(x) x[!is.na(x)])
+
+  # Remove non-interactions (same variable)
+  interaction_list <- lapply(interaction_list, unique)  # remove same variables
+  interaction_length <- sapply(interaction_list, length)
+  interaction_list <- interaction_list[interaction_length > 1]
+  interaction_list <- unique(lapply(interaction_list, sort))
+  return(interaction_list)
+}
+
+# Generate sample data
+x <- list()
+for (i in 1:10){
+  x[[i]] = i*rnorm(1000, 10)
+}
+x <- as.data.table(x)
+
+y = -1*x[, rowSums(.SD)] + x[['V1']]*x[['V2']] + x[['V3']]*x[['V4']]*x[['V5']] + rnorm(1000, 0.001) + 3*sin(x[['V7']])
+
+train = as.matrix(x)
+
+# Interaction constraint list (column names form)
+interaction_list <- list(c('V1','V2'),c('V3','V4','V5'))
+
+# Convert interaction constraint list into feature index form
+cols2ids <- function(object, col_names) {
+  LUT <- seq_along(col_names) - 1
+  names(LUT) <- col_names
+  rapply(object, function(x) LUT[x], classes="character", how="replace")
+}
+interaction_list_fid = cols2ids(interaction_list, colnames(train))
+
+# Fit model with interaction constraints
+bst = xgboost(data = train, label = y, max_depth = 4,
+              eta = 0.1, nthread = 2, nrounds = 1000,
+              interaction_constraints = interaction_list_fid)
+
+bst_tree <- xgb.model.dt.tree(colnames(train), bst)
+bst_interactions <- treeInteractions(bst_tree, 4)  # interactions constrained to combinations of V1*V2 and V3*V4*V5
+
+# Fit model without interaction constraints
+bst2 = xgboost(data = train, label = y, max_depth = 4,
+               eta = 0.1, nthread = 2, nrounds = 1000)
+
+bst2_tree <- xgb.model.dt.tree(colnames(train), bst2)
+bst2_interactions <- treeInteractions(bst2_tree, 4)  # much more interactions
+
+# Fit model with both interaction and monotonicity constraints
+bst3 = xgboost(data = train, label = y, max_depth = 4,
+               eta = 0.1, nthread = 2, nrounds = 1000,
+               interaction_constraints = interaction_list_fid,
+               monotone_constraints = c(-1,0,0,0,0,0,0,0,0,0))
+
+bst3_tree <- xgb.model.dt.tree(colnames(train), bst3)
+bst3_interactions <- treeInteractions(bst3_tree, 4)  # interactions still constrained to combinations of V1*V2 and V3*V4*V5
+
+# Show monotonic constraints still apply by checking scores after incrementing V1
+x1 <- sort(unique(x[['V1']]))
+for (i in 1:length(x1)){
+  testdata <- copy(x[, -c('V1')])
+  testdata[['V1']] <- x1[i]
+  testdata <- testdata[, paste0('V',1:10), with=F]
+  pred <- predict(bst3, as.matrix(testdata))
+  
+  # Should not print out anything due to monotonic constraints
+  if (i > 1) if (any(pred > prev_pred)) print(i)
+  prev_pred <- pred 
+}

R-package/demo/predict_first_ntree.R

@@ -7,10 +7,10 @@ dtest <- xgb.DMatrix(agaricus.test$data, label = agaricus.test$label)
 
 param <- list(max_depth=2, eta=1, silent=1, objective='binary:logistic')
 watchlist <- list(eval = dtest, train = dtrain)
-nround = 2
+nrounds = 2
 
 # training the model for two rounds
-bst = xgb.train(param, dtrain, nround, nthread = 2, watchlist)
+bst = xgb.train(param, dtrain, nrounds, nthread = 2, watchlist)
 cat('start testing prediction from first n trees\n')
 labels <- getinfo(dtest,'label')
 

R-package/demo/predict_leaf_indices.R

@@ -11,10 +11,10 @@ dtrain <- xgb.DMatrix(data = agaricus.train$data, label = agaricus.train$label)
 dtest <- xgb.DMatrix(data = agaricus.test$data, label = agaricus.test$label)
 
 param <- list(max_depth=2, eta=1, silent=1, objective='binary:logistic')
-nround = 4
+nrounds = 4
 
 # training the model for two rounds
-bst = xgb.train(params = param, data = dtrain, nrounds = nround, nthread = 2)
+bst = xgb.train(params = param, data = dtrain, nrounds = nrounds, nthread = 2)
 
 # Model accuracy without new features
 accuracy.before <- sum((predict(bst, agaricus.test$data) >= 0.5) == agaricus.test$label) / length(agaricus.test$label)
@@ -38,12 +38,13 @@ create.new.tree.features <- function(model, original.features){
 # Convert previous features to one hot encoding
 new.features.train <- create.new.tree.features(bst, agaricus.train$data)
 new.features.test <- create.new.tree.features(bst, agaricus.test$data)
+colnames(new.features.test) <- colnames(new.features.train)
 
 # learning with new features
 new.dtrain <- xgb.DMatrix(data = new.features.train, label = agaricus.train$label)
 new.dtest <- xgb.DMatrix(data = new.features.test, label = agaricus.test$label)
 watchlist <- list(train = new.dtrain)
-bst <- xgb.train(params = param, data = new.dtrain, nrounds = nround, nthread = 2)
+bst <- xgb.train(params = param, data = new.dtrain, nrounds = nrounds, nthread = 2)
 
 # Model accuracy with new features
 accuracy.after <- sum((predict(bst, new.dtest) >= 0.5) == agaricus.test$label) / length(agaricus.test$label)

R-package/man/callbacks.Rd

@@ -5,24 +5,24 @@
 \title{Callback closures for booster training.}
 \description{
 These are used to perform various service tasks either during boosting iterations or at the end.
-This approach helps to modularize many of such tasks without bloating the main training methods, 
+This approach helps to modularize many of such tasks without bloating the main training methods,
 and it offers .
 }
 \details{
 By default, a callback function is run after each boosting iteration.
 An R-attribute \code{is_pre_iteration} could be set for a callback to define a pre-iteration function.
 
-When a callback function has \code{finalize} parameter, its finalizer part will also be run after 
+When a callback function has \code{finalize} parameter, its finalizer part will also be run after
 the boosting is completed.
 
-WARNING: side-effects!!! Be aware that these callback functions access and modify things in 
+WARNING: side-effects!!! Be aware that these callback functions access and modify things in
 the environment from which they are called from, which is a fairly uncommon thing to do in R.
 
-To write a custom callback closure, make sure you first understand the main concepts about R envoronments.
-Check either R documentation on \code{\link[base]{environment}} or the 
-\href{http://adv-r.had.co.nz/Environments.html}{Environments chapter} from the "Advanced R" 
+To write a custom callback closure, make sure you first understand the main concepts about R environments.
+Check either R documentation on \code{\link[base]{environment}} or the
+\href{http://adv-r.had.co.nz/Environments.html}{Environments chapter} from the "Advanced R"
 book by Hadley Wickham. Further, the best option is to read the code of some of the existing callbacks -
-choose ones that do something similar to what you want to achieve. Also, you would need to get familiar 
+choose ones that do something similar to what you want to achieve. Also, you would need to get familiar
 with the objects available inside of the \code{xgb.train} and \code{xgb.cv} internal environments.
 }
 \seealso{

R-package/man/cb.cv.predict.Rd

@@ -11,11 +11,11 @@ cb.cv.predict(save_models = FALSE)
 }
 \value{
 Predictions are returned inside of the \code{pred} element, which is either a vector or a matrix,
-depending on the number of prediction outputs per data row. The order of predictions corresponds 
-to the order of rows in the original dataset. Note that when a custom \code{folds} list is 
-provided in \code{xgb.cv}, the predictions would only be returned properly when this list is a 
-non-overlapping list of k sets of indices, as in a standard k-fold CV. The predictions would not be 
-meaningful when user-profided folds have overlapping indices as in, e.g., random sampling splits.
+depending on the number of prediction outputs per data row. The order of predictions corresponds
+to the order of rows in the original dataset. Note that when a custom \code{folds} list is
+provided in \code{xgb.cv}, the predictions would only be returned properly when this list is a
+non-overlapping list of k sets of indices, as in a standard k-fold CV. The predictions would not be
+meaningful when user-provided folds have overlapping indices as in, e.g., random sampling splits.
 When some of the indices in the training dataset are not included into user-provided \code{folds},
 their prediction value would be \code{NA}.
 }

R-package/man/cb.early.stop.Rd

@@ -8,15 +8,15 @@ cb.early.stop(stopping_rounds, maximize = FALSE, metric_name = NULL,
   verbose = TRUE)
 }
 \arguments{
-\item{stopping_rounds}{The number of rounds with no improvement in 
+\item{stopping_rounds}{The number of rounds with no improvement in
 the evaluation metric in order to stop the training.}
 
 \item{maximize}{whether to maximize the evaluation metric}
 
 \item{metric_name}{the name of an evaluation column to use as a criteria for early
 stopping. If not set, the last column would be used.
-Let's say the test data in \code{watchlist} was labelled as \code{dtest}, 
-and one wants to use the AUC in test data for early stopping regardless of where 
+Let's say the test data in \code{watchlist} was labelled as \code{dtest},
+and one wants to use the AUC in test data for early stopping regardless of where
 it is in the \code{watchlist}, then one of the following would need to be set:
 \code{metric_name='dtest-auc'} or \code{metric_name='dtest_auc'}.
 All dash '-' characters in metric names are considered equivalent to '_'.}
@@ -27,7 +27,7 @@ All dash '-' characters in metric names are considered equivalent to '_'.}
 Callback closure to activate the early stopping.
 }
 \details{
-This callback function determines the condition for early stopping 
+This callback function determines the condition for early stopping
 by setting the \code{stop_condition = TRUE} flag in its calling frame.
 
 The following additional fields are assigned to the model's R object:

R-package/man/cb.evaluation.log.Rd

@@ -13,12 +13,12 @@ Callback closure for logging the evaluation history
 This callback function appends the current iteration evaluation results \code{bst_evaluation}
 available in the calling parent frame to the \code{evaluation_log} list in a calling frame.
 
-The finalizer callback (called with \code{finalize = TURE} in the end) converts 
+The finalizer callback (called with \code{finalize = TURE} in the end) converts
 the \code{evaluation_log} list into a final data.table.
 
-The iteration evaluation result \code{bst_evaluation} must be a named numeric vector. 
+The iteration evaluation result \code{bst_evaluation} must be a named numeric vector.
 
-Note: in the column names of the final data.table, the dash '-' character is replaced with 
+Note: in the column names of the final data.table, the dash '-' character is replaced with
 the underscore '_' in order to make the column names more like regular R identifiers.
 
 Callback function expects the following values to be set in its calling frame:

R-package/man/cb.reset.parameters.Rd

@@ -2,27 +2,27 @@
 % Please edit documentation in R/callbacks.R
 \name{cb.reset.parameters}
 \alias{cb.reset.parameters}
-\title{Callback closure for restetting the booster's parameters at each iteration.}
+\title{Callback closure for resetting the booster's parameters at each iteration.}
 \usage{
 cb.reset.parameters(new_params)
 }
 \arguments{
 \item{new_params}{a list where each element corresponds to a parameter that needs to be reset.
-Each element's value must be either a vector of values of length \code{nrounds} 
-to be set at each iteration, 
-or a function of two parameters \code{learning_rates(iteration, nrounds)} 
-which returns a new parameter value by using the current iteration number 
+Each element's value must be either a vector of values of length \code{nrounds}
+to be set at each iteration,
+or a function of two parameters \code{learning_rates(iteration, nrounds)}
+which returns a new parameter value by using the current iteration number
 and the total number of boosting rounds.}
 }
 \description{
-Callback closure for restetting the booster's parameters at each iteration.
+Callback closure for resetting the booster's parameters at each iteration.
 }
 \details{
 This is a "pre-iteration" callback function used to reset booster's parameters
 at the beginning of each iteration.
 
-Note that when training is resumed from some previous model, and a function is used to 
-reset a parameter value, the \code{nround} argument in this function would be the 
+Note that when training is resumed from some previous model, and a function is used to
+reset a parameter value, the \code{nrounds} argument in this function would be the
 the number of boosting rounds in the current training.
 
 Callback function expects the following values to be set in its calling frame:

R-package/man/cb.save.model.Rd

@@ -7,13 +7,13 @@
 cb.save.model(save_period = 0, save_name = "xgboost.model")
 }
 \arguments{
-\item{save_period}{save the model to disk after every 
+\item{save_period}{save the model to disk after every
 \code{save_period} iterations; 0 means save the model at the end.}
 
 \item{save_name}{the name or path for the saved model file.
-It can contain a \code{\link[base]{sprintf}} formatting specifier 
+It can contain a \code{\link[base]{sprintf}} formatting specifier
 to include the integer iteration number in the file name.
-E.g., with \code{save_name} = 'xgboost_%04d.model', 
+E.g., with \code{save_name} = 'xgboost_%04d.model',
 the file saved at iteration 50 would be named "xgboost_0050.model".}
 }
 \description{

R-package/man/dim.xgb.DMatrix.Rd

@@ -13,7 +13,7 @@
 Returns a vector of numbers of rows and of columns in an \code{xgb.DMatrix}.
 }
 \details{
-Note: since \code{nrow} and \code{ncol} internally use \code{dim}, they can also 
+Note: since \code{nrow} and \code{ncol} internally use \code{dim}, they can also
 be directly used with an \code{xgb.DMatrix} object.
 }
 \examples{

R-package/man/dimnames.xgb.DMatrix.Rd

@@ -16,8 +16,8 @@
 and the second one is column names}
 }
 \description{
-Only column names are supported for \code{xgb.DMatrix}, thus setting of 
-row names would have no effect and returnten row names would be NULL.
+Only column names are supported for \code{xgb.DMatrix}, thus setting of
+row names would have no effect and returned row names would be NULL.
 }
 \details{
 Generic \code{dimnames} methods are used by \code{colnames}.

R-package/man/getinfo.Rd

@@ -27,7 +27,7 @@ The \code{name} field can be one of the following:
     \item \code{weight}: to do a weight rescale ;
     \item \code{base_margin}: base margin is the base prediction Xgboost will boost from ;
     \item \code{nrow}: number of rows of the \code{xgb.DMatrix}.
-    
+
 }
 
 \code{group} can be setup by \code{setinfo} but can't be retrieved by \code{getinfo}.

R-package/man/predict.xgb.Booster.Rd

@@ -7,7 +7,8 @@
 \usage{
 \method{predict}{xgb.Booster}(object, newdata, missing = NA,
   outputmargin = FALSE, ntreelimit = NULL, predleaf = FALSE,
-  predcontrib = FALSE, approxcontrib = FALSE, reshape = FALSE, ...)
+  predcontrib = FALSE, approxcontrib = FALSE, predinteraction = FALSE,
+  reshape = FALSE, ...)
 
 \method{predict}{xgb.Booster.handle}(object, ...)
 }
@@ -26,14 +27,17 @@ logistic regression would result in predictions for log-odds instead of probabil
 \item{ntreelimit}{limit the number of model's trees or boosting iterations used in prediction (see Details).
 It will use all the trees by default (\code{NULL} value).}
 
-\item{predleaf}{whether predict leaf index instead.}
+\item{predleaf}{whether predict leaf index.}
 
-\item{predcontrib}{whether to return feature contributions to individual predictions instead (see Details).}
+\item{predcontrib}{whether to return feature contributions to individual predictions (see Details).}
 
 \item{approxcontrib}{whether to use a fast approximation for feature contributions (see Details).}
 
+\item{predinteraction}{whether to return contributions of feature interactions to individual predictions (see Details).}
+
 \item{reshape}{whether to reshape the vector of predictions to a matrix form when there are several
-prediction outputs per case. This option has no effect when \code{predleaf = TRUE}.}
+prediction outputs per case. This option has no effect when either of predleaf, predcontrib,
+or predinteraction flags is TRUE.}
 
 \item{...}{Parameters passed to \code{predict.xgb.Booster}}
 }
@@ -51,6 +55,14 @@ When \code{predcontrib = TRUE} and it is not a multiclass setting, the output is
 For a multiclass case, a list of \code{num_class} elements is returned, where each element is
 such a matrix. The contribution values are on the scale of untransformed margin
 (e.g., for binary classification would mean that the contributions are log-odds deviations from bias).
+
+When \code{predinteraction = TRUE} and it is not a multiclass setting, the output is a 3d array with
+dimensions \code{c(nrow, num_features + 1, num_features + 1)}. The off-diagonal (in the last two dimensions)
+elements represent different features interaction contributions. The array is symmetric WRT the last
+two dimensions. The "+ 1" columns corresponds to bias. Summing this array along the last dimension should
+produce practically the same result as predict with \code{predcontrib = TRUE}.
+For a multiclass case, a list of \code{num_class} elements is returned, where each element is
+such an array.
 }
 \description{
 Predicted values based on either xgboost model or model handle object.
@@ -76,6 +88,11 @@ values (Lundberg 2017) that sum to the difference between the expected output
 of the model and the current prediction (where the hessian weights are used to compute the expectations).
 Setting \code{approxcontrib = TRUE} approximates these values following the idea explained
 in \url{http://blog.datadive.net/interpreting-random-forests/}.
+
+With \code{predinteraction = TRUE}, SHAP values of contributions of interaction of each pair of features
+are computed. Note that this operation might be rather expensive in terms of compute and memory.
+Since it quadratically depends on the number of features, it is recommended to perform selection
+of the most important features first. See below about the format of the returned results.
 }
 \examples{
 ## binary classification:

R-package/man/print.xgb.DMatrix.Rd

@@ -14,7 +14,7 @@
 \item{...}{not currently used}
 }
 \description{
-Print information about xgb.DMatrix. 
+Print information about xgb.DMatrix.
 Currently it displays dimensions and presence of info-fields and colnames.
 }
 \examples{

R-package/man/print.xgb.cv.Rd

@@ -17,7 +17,7 @@
 Prints formatted results of \code{xgb.cv}.
 }
 \details{
-When not verbose, it would only print the evaluation results, 
+When not verbose, it would only print the evaluation results,
 including the best iteration (when available).
 }
 \examples{

R-package/man/slice.xgb.DMatrix.Rd

@@ -5,7 +5,7 @@
 \alias{slice.xgb.DMatrix}
 \alias{[.xgb.DMatrix}
 \title{Get a new DMatrix containing the specified rows of
-orginal xgb.DMatrix object}
+original xgb.DMatrix object}
 \usage{
 slice(object, ...)
 
@@ -24,7 +24,7 @@ slice(object, ...)
 }
 \description{
 Get a new DMatrix containing the specified rows of
-orginal xgb.DMatrix object
+original xgb.DMatrix object
 }
 \examples{
 data(agaricus.train, package='xgboost')

R-package/man/xgb.Booster.complete.Rd

@@ -28,7 +28,7 @@ E.g., when an \code{xgb.Booster} model is saved as an R object and then is loade
 its handle (pointer) to an internal xgboost model would be invalid. The majority of xgboost methods
 should still work for such a model object since those methods would be using
 \code{xgb.Booster.complete} internally. However, one might find it to be more efficient to call the
-\code{xgb.Booster.complete} function explicitely once after loading a model as an R-object.
+\code{xgb.Booster.complete} function explicitly once after loading a model as an R-object.
 That would prevent further repeated implicit reconstruction of an internal booster model.
 }
 \examples{
@@ -39,6 +39,7 @@ bst <- xgboost(data = agaricus.train$data, label = agaricus.train$label, max_dep
 saveRDS(bst, "xgb.model.rds")
 
 bst1 <- readRDS("xgb.model.rds")
+if (file.exists("xgb.model.rds")) file.remove("xgb.model.rds")
 # the handle is invalid:
 print(bst1$handle)
 

R-package/man/xgb.DMatrix.Rd

@@ -7,7 +7,7 @@
 xgb.DMatrix(data, info = list(), missing = NA, silent = FALSE, ...)
 }
 \arguments{
-\item{data}{a \code{matrix} object (either numeric or integer), a \code{dgCMatrix} object, or a character 
+\item{data}{a \code{matrix} object (either numeric or integer), a \code{dgCMatrix} object, or a character
 string representing a filename.}
 
 \item{info}{a named list of additional information to store in the \code{xgb.DMatrix} object.
@@ -31,4 +31,5 @@ train <- agaricus.train
 dtrain <- xgb.DMatrix(train$data, label=train$label)
 xgb.DMatrix.save(dtrain, 'xgb.DMatrix.data')
 dtrain <- xgb.DMatrix('xgb.DMatrix.data')
+if (file.exists('xgb.DMatrix.data')) file.remove('xgb.DMatrix.data')
 }

R-package/man/xgb.DMatrix.save.Rd

@@ -20,4 +20,5 @@ train <- agaricus.train
 dtrain <- xgb.DMatrix(train$data, label=train$label)
 xgb.DMatrix.save(dtrain, 'xgb.DMatrix.data')
 dtrain <- xgb.DMatrix('xgb.DMatrix.data')
+if (file.exists('xgb.DMatrix.data')) file.remove('xgb.DMatrix.data')
 }

R-package/man/xgb.attr.Rd

@@ -73,6 +73,7 @@ xgb.attributes(bst) <- list(a = 123, b = "abc")
 
 xgb.save(bst, 'xgb.model')
 bst1 <- xgb.load('xgb.model')
+if (file.exists('xgb.model')) file.remove('xgb.model')
 print(xgb.attr(bst1, "my_attribute"))
 print(xgb.attributes(bst1))
 

R-package/man/xgb.create.features.Rd

@@ -63,9 +63,9 @@ dtrain <- xgb.DMatrix(data = agaricus.train$data, label = agaricus.train$label)
 dtest <- xgb.DMatrix(data = agaricus.test$data, label = agaricus.test$label)
 
 param <- list(max_depth=2, eta=1, silent=1, objective='binary:logistic')
-nround = 4
+nrounds = 4
 
-bst = xgb.train(params = param, data = dtrain, nrounds = nround, nthread = 2)
+bst = xgb.train(params = param, data = dtrain, nrounds = nrounds, nthread = 2)
 
 # Model accuracy without new features
 accuracy.before <- sum((predict(bst, agaricus.test$data) >= 0.5) == agaricus.test$label) /
@@ -79,7 +79,7 @@ new.features.test <- xgb.create.features(model = bst, agaricus.test$data)
 new.dtrain <- xgb.DMatrix(data = new.features.train, label = agaricus.train$label)
 new.dtest <- xgb.DMatrix(data = new.features.test, label = agaricus.test$label)
 watchlist <- list(train = new.dtrain)
-bst <- xgb.train(params = param, data = new.dtrain, nrounds = nround, nthread = 2)
+bst <- xgb.train(params = param, data = new.dtrain, nrounds = nrounds, nthread = 2)
 
 # Model accuracy with new features
 accuracy.after <- sum((predict(bst, new.dtest) >= 0.5) == agaricus.test$label) /

R-package/man/xgb.cv.Rd

@@ -15,7 +15,7 @@ xgb.cv(params = list(), data, nrounds, nfold, label = NULL, missing = NA,
 \itemize{
   \item \code{objective} objective function, common ones are
   \itemize{
-    \item \code{reg:linear} linear regression
+    \item \code{reg:squarederror} Regression with squared loss
     \item \code{binary:logistic} logistic regression for classification
   }
   \item \code{eta} step size of each boosting step
@@ -34,11 +34,11 @@ xgb.cv(params = list(), data, nrounds, nfold, label = NULL, missing = NA,
 
 \item{label}{vector of response values. Should be provided only when data is an R-matrix.}
 
-\item{missing}{is only used when input is a dense matrix. By default is set to NA, which means 
-that NA values should be considered as 'missing' by the algorithm. 
+\item{missing}{is only used when input is a dense matrix. By default is set to NA, which means
+that NA values should be considered as 'missing' by the algorithm.
 Sometimes, 0 or other extreme value might be used to represent missing values.}
 
-\item{prediction}{A logical value indicating whether to return the test fold predictions 
+\item{prediction}{A logical value indicating whether to return the test fold predictions
 from each CV model. This parameter engages the \code{\link{cb.cv.predict}} callback.}
 
 \item{showsd}{\code{boolean}, whether to show standard deviation of cross validation}
@@ -55,28 +55,28 @@ from each CV model. This parameter engages the \code{\link{cb.cv.predict}} callb
   \item \code{merror} Exact matching error, used to evaluate multi-class classification
 }}
 
-\item{obj}{customized objective function. Returns gradient and second order 
+\item{obj}{customized objective function. Returns gradient and second order
 gradient with given prediction and dtrain.}
 
-\item{feval}{custimized evaluation function. Returns 
-\code{list(metric='metric-name', value='metric-value')} with given 
+\item{feval}{customized evaluation function. Returns
+\code{list(metric='metric-name', value='metric-value')} with given
 prediction and dtrain.}
 
-\item{stratified}{a \code{boolean} indicating whether sampling of folds should be stratified 
+\item{stratified}{a \code{boolean} indicating whether sampling of folds should be stratified
 by the values of outcome labels.}
 
 \item{folds}{\code{list} provides a possibility to use a list of pre-defined CV folds
-(each element must be a vector of test fold's indices). When folds are supplied, 
+(each element must be a vector of test fold's indices). When folds are supplied,
 the \code{nfold} and \code{stratified} parameters are ignored.}
 
 \item{verbose}{\code{boolean}, print the statistics during the process}
 
 \item{print_every_n}{Print each n-th iteration evaluation messages when \code{verbose>0}.
-Default is 1 which means all messages are printed. This parameter is passed to the 
+Default is 1 which means all messages are printed. This parameter is passed to the
 \code{\link{cb.print.evaluation}} callback.}
 
-\item{early_stopping_rounds}{If \code{NULL}, the early stopping function is not triggered. 
-If set to an integer \code{k}, training with a validation set will stop if the performance 
+\item{early_stopping_rounds}{If \code{NULL}, the early stopping function is not triggered.
+If set to an integer \code{k}, training with a validation set will stop if the performance
 doesn't improve for \code{k} rounds.
 Setting this parameter engages the \code{\link{cb.early.stop}} callback.}
 
@@ -86,8 +86,8 @@ When it is \code{TRUE}, it means the larger the evaluation score the better.
 This parameter is passed to the \code{\link{cb.early.stop}} callback.}
 
 \item{callbacks}{a list of callback functions to perform various task during boosting.
-See \code{\link{callbacks}}. Some of the callbacks are automatically created depending on the 
-parameters' values. User can provide either existing or their own callback methods in order 
+See \code{\link{callbacks}}. Some of the callbacks are automatically created depending on the
+parameters' values. User can provide either existing or their own callback methods in order
 to customize the training process.}
 
 \item{...}{other parameters to pass to \code{params}.}
@@ -96,26 +96,26 @@ to customize the training process.}
 An object of class \code{xgb.cv.synchronous} with the following elements:
 \itemize{
   \item \code{call} a function call.
-  \item \code{params} parameters that were passed to the xgboost library. Note that it does not 
+  \item \code{params} parameters that were passed to the xgboost library. Note that it does not
         capture parameters changed by the \code{\link{cb.reset.parameters}} callback.
-  \item \code{callbacks} callback functions that were either automatically assigned or 
+  \item \code{callbacks} callback functions that were either automatically assigned or
         explicitly passed.
-  \item \code{evaluation_log} evaluation history storead as a \code{data.table} with the
-        first column corresponding to iteration number and the rest corresponding to the 
+  \item \code{evaluation_log} evaluation history stored as a \code{data.table} with the
+        first column corresponding to iteration number and the rest corresponding to the
         CV-based evaluation means and standard deviations for the training and test CV-sets.
         It is created by the \code{\link{cb.evaluation.log}} callback.
   \item \code{niter} number of boosting iterations.
   \item \code{nfeatures} number of features in training data.
-  \item \code{folds} the list of CV folds' indices - either those passed through the \code{folds} 
+  \item \code{folds} the list of CV folds' indices - either those passed through the \code{folds}
         parameter or randomly generated.
   \item \code{best_iteration} iteration number with the best evaluation metric value
         (only available with early stopping).
-  \item \code{best_ntreelimit} the \code{ntreelimit} value corresponding to the best iteration, 
+  \item \code{best_ntreelimit} the \code{ntreelimit} value corresponding to the best iteration,
         which could further be used in \code{predict} method
         (only available with early stopping).
-  \item \code{pred} CV prediction values available when \code{prediction} is set. 
+  \item \code{pred} CV prediction values available when \code{prediction} is set.
         It is either vector or matrix (see \code{\link{cb.cv.predict}}).
-  \item \code{models} a liost of the CV folds' models. It is only available with the explicit 
+  \item \code{models} a liost of the CV folds' models. It is only available with the explicit
         setting of the \code{cb.cv.predict(save_models = TRUE)} callback.
 }
 }
@@ -123,9 +123,9 @@ An object of class \code{xgb.cv.synchronous} with the following elements:
 The cross validation function of xgboost
 }
 \details{
-The original sample is randomly partitioned into \code{nfold} equal size subsamples. 
+The original sample is randomly partitioned into \code{nfold} equal size subsamples.
 
-Of the \code{nfold} subsamples, a single subsample is retained as the validation data for testing the model, and the remaining \code{nfold - 1} subsamples are used as training data. 
+Of the \code{nfold} subsamples, a single subsample is retained as the validation data for testing the model, and the remaining \code{nfold - 1} subsamples are used as training data.
 
 The cross-validation process is then repeated \code{nrounds} times, with each of the \code{nfold} subsamples used exactly once as the validation data.
 

R-package/man/xgb.dump.Rd

@@ -44,8 +44,8 @@ test <- agaricus.test
 bst <- xgboost(data = train$data, label = train$label, max_depth = 2, 
                eta = 1, nthread = 2, nrounds = 2, objective = "binary:logistic")
 # save the model in file 'xgb.model.dump'
-dump.path = file.path(tempdir(), 'model.dump')
-xgb.dump(bst, dump.path, with_stats = TRUE)
+dump_path = file.path(tempdir(), 'model.dump')
+xgb.dump(bst, dump_path, with_stats = TRUE)
 
 # print the model without saving it to a file
 print(xgb.dump(bst, with_stats = TRUE))

R-package/man/xgb.gblinear.history.Rd

@@ -12,7 +12,7 @@ using the \code{cb.gblinear.history()} callback.}
 
 \item{class_index}{zero-based class index to extract the coefficients for only that
 specific class in a multinomial multiclass model. When it is NULL, all the
-coeffients are returned. Has no effect in non-multiclass models.}
+coefficients are returned. Has no effect in non-multiclass models.}
 }
 \value{
 For an \code{xgb.train} result, a matrix (either dense or sparse) with the columns

R-package/man/xgb.load.Rd

@@ -33,6 +33,7 @@ bst <- xgboost(data = train$data, label = train$label, max_depth = 2,
                eta = 1, nthread = 2, nrounds = 2,objective = "binary:logistic")
 xgb.save(bst, 'xgb.model')
 bst <- xgb.load('xgb.model')
+if (file.exists('xgb.model')) file.remove('xgb.model')
 pred <- predict(bst, test$data)
 }
 \seealso{

R-package/man/xgb.plot.deepness.Rd

@@ -50,7 +50,7 @@ per tree with respect to tree number are created. And \code{which="med.weight"}
 a tree's median absolute leaf weight changes through the iterations.
 
 This function was inspired by the blog post
-\url{http://aysent.github.io/2015/11/08/random-forest-leaf-visualization.html}.
+\url{https://github.com/aysent/random-forest-leaf-visualization}.
 }
 \examples{
 

R-package/man/xgb.plot.importance.Rd

@@ -17,13 +17,13 @@ xgb.plot.importance(importance_matrix = NULL, top_n = NULL,
 
 \item{top_n}{maximal number of top features to include into the plot.}
 
-\item{measure}{the name of importance measure to plot. 
+\item{measure}{the name of importance measure to plot.
 When \code{NULL}, 'Gain' would be used for trees and 'Weight' would be used for gblinear.}
 
 \item{rel_to_first}{whether importance values should be represented as relative to the highest ranked feature.
 See Details.}
 
-\item{n_clusters}{(ggplot only) a \code{numeric} vector containing the min and the max range 
+\item{n_clusters}{(ggplot only) a \code{numeric} vector containing the min and the max range
 of the possible number of clusters of bars.}
 
 \item{...}{other parameters passed to \code{barplot} (except horiz, border, cex.names, names.arg, and las).}
@@ -33,7 +33,7 @@ When it is NULL, the existing \code{par('mar')} is used.}
 
 \item{cex}{(base R barplot) passed as \code{cex.names} parameter to \code{barplot}.}
 
-\item{plot}{(base R barplot) whether a barplot should be produced. 
+\item{plot}{(base R barplot) whether a barplot should be produced.
 If FALSE, only a data.table is returned.}
 }
 \value{
@@ -53,14 +53,14 @@ Features are shown ranked in a decreasing importance order.
 It works for importances from both \code{gblinear} and \code{gbtree} models.
 
 When \code{rel_to_first = FALSE}, the values would be plotted as they were in \code{importance_matrix}.
-For gbtree model, that would mean being normalized to the total of 1 
+For gbtree model, that would mean being normalized to the total of 1
 ("what is feature's importance contribution relative to the whole model?").
 For linear models, \code{rel_to_first = FALSE} would show actual values of the coefficients.
-Setting \code{rel_to_first = TRUE} allows to see the picture from the perspective of 
+Setting \code{rel_to_first = TRUE} allows to see the picture from the perspective of
 "what is feature's importance contribution relative to the most important feature?"
 
-The ggplot-backend method also performs 1-D custering of the importance values, 
-with bar colors coresponding to different clusters that have somewhat similar importance values.
+The ggplot-backend method also performs 1-D clustering of the importance values,
+with bar colors corresponding to different clusters that have somewhat similar importance values.
 }
 \examples{
 data(agaricus.train)

R-package/man/xgb.plot.shap.Rd

@@ -15,7 +15,7 @@ xgb.plot.shap(data, shap_contrib = NULL, features = NULL, top_n = 1,
 \arguments{
 \item{data}{data as a \code{matrix} or \code{dgCMatrix}.}
 
-\item{shap_contrib}{a matrix of SHAP contributions that was computed earlier for the above 
+\item{shap_contrib}{a matrix of SHAP contributions that was computed earlier for the above
 \code{data}. When it is NULL, it is computed internally using \code{model} and \code{data}.}
 
 \item{features}{a vector of either column indices or of feature names to plot. When it is NULL,
@@ -63,7 +63,7 @@ more than 5 distinct values.}
 
 \item{col_loess}{a color to use for the loess curves.}
 
-\item{span_loess}{the \code{span} paramerer in \code{\link[stats]{loess}}'s call.}
+\item{span_loess}{the \code{span} parameter in \code{\link[stats]{loess}}'s call.}
 
 \item{which}{whether to do univariate or bivariate plotting. NOTE: only 1D is implemented so far.}
 
@@ -104,7 +104,7 @@ a meaningful thing to do.
 data(agaricus.train, package='xgboost')
 data(agaricus.test, package='xgboost')
 
-bst <- xgboost(agaricus.train$data, agaricus.train$label, nrounds = 50, 
+bst <- xgboost(agaricus.train$data, agaricus.train$label, nrounds = 50,
                eta = 0.1, max_depth = 3, subsample = .5,
                method = "hist", objective = "binary:logistic", nthread = 2, verbose = 0)
 

R-package/man/xgb.save.Rd

@@ -33,6 +33,7 @@ bst <- xgboost(data = train$data, label = train$label, max_depth = 2,
                eta = 1, nthread = 2, nrounds = 2,objective = "binary:logistic")
 xgb.save(bst, 'xgb.model')
 bst <- xgb.load('xgb.model')
+if (file.exists('xgb.model')) file.remove('xgb.model')
 pred <- predict(bst, test$data)
 }
 \seealso{

R-package/man/xgb.train.Rd

@@ -16,7 +16,7 @@ xgboost(data = NULL, label = NULL, missing = NA, weight = NULL,
   save_name = "xgboost.model", xgb_model = NULL, callbacks = list(), ...)
 }
 \arguments{
-\item{params}{the list of parameters. 
+\item{params}{the list of parameters.
        The complete list of parameters is available at \url{http://xgboost.readthedocs.io/en/latest/parameter.html}.
        Below is a shorter summary:
 
@@ -25,36 +25,37 @@ xgboost(data = NULL, label = NULL, missing = NA, weight = NULL,
 \itemize{
   \item \code{booster} which booster to use, can be \code{gbtree} or \code{gblinear}. Default: \code{gbtree}.
 }
- 
+
 2. Booster Parameters
 
 2.1. Parameter for Tree Booster
 
 \itemize{
   \item \code{eta} control the learning rate: scale the contribution of each tree by a factor of \code{0 < eta < 1} when it is added to the current approximation. Used to prevent overfitting by making the boosting process more conservative. Lower value for \code{eta} implies larger value for \code{nrounds}: low \code{eta} value means model more robust to overfitting but slower to compute. Default: 0.3
-  \item \code{gamma} minimum loss reduction required to make a further partition on a leaf node of the tree. the larger, the more conservative the algorithm will be. 
+  \item \code{gamma} minimum loss reduction required to make a further partition on a leaf node of the tree. the larger, the more conservative the algorithm will be.
   \item \code{max_depth} maximum depth of a tree. Default: 6
   \item \code{min_child_weight} minimum sum of instance weight (hessian) needed in a child. If the tree partition step results in a leaf node with the sum of instance weight less than min_child_weight, then the building process will give up further partitioning. In linear regression mode, this simply corresponds to minimum number of instances needed to be in each node. The larger, the more conservative the algorithm will be. Default: 1
-  \item \code{subsample} subsample ratio of the training instance. Setting it to 0.5 means that xgboost randomly collected half of the data instances to grow trees and this will prevent overfitting. It makes computation shorter (because less data to analyse). It is advised to use this parameter with \code{eta} and increase \code{nround}. Default: 1 
+  \item \code{subsample} subsample ratio of the training instance. Setting it to 0.5 means that xgboost randomly collected half of the data instances to grow trees and this will prevent overfitting. It makes computation shorter (because less data to analyse). It is advised to use this parameter with \code{eta} and increase \code{nrounds}. Default: 1
   \item \code{colsample_bytree} subsample ratio of columns when constructing each tree. Default: 1
   \item \code{num_parallel_tree} Experimental parameter. number of trees to grow per round. Useful to test Random Forest through Xgboost (set \code{colsample_bytree < 1}, \code{subsample  < 1}  and \code{round = 1}) accordingly. Default: 1
   \item \code{monotone_constraints} A numerical vector consists of \code{1}, \code{0} and \code{-1} with its length equals to the number of features in the training data. \code{1} is increasing, \code{-1} is decreasing and \code{0} is no constraint.
+  \item \code{interaction_constraints} A list of vectors specifying feature indices of permitted interactions. Each item of the list represents one permitted interaction where specified features are allowed to interact with each other. Feature index values should start from \code{0} (\code{0} references the first column).  Leave argument unspecified for no interaction constraints.
 }
 
 2.2. Parameter for Linear Booster
- 
+
 \itemize{
   \item \code{lambda} L2 regularization term on weights. Default: 0
   \item \code{lambda_bias} L2 regularization term on bias. Default: 0
   \item \code{alpha} L1 regularization term on weights. (there is no L1 reg on bias because it is not important). Default: 0
 }
 
-3. Task Parameters 
+3. Task Parameters
 
 \itemize{
 \item \code{objective} specify the learning task and the corresponding learning objective, users can pass a self-defined function to it. The default objective options are below:
   \itemize{
-    \item \code{reg:linear} linear regression (Default).
+    \item \code{reg:squarederror} Regression with squared loss (Default).
     \item \code{reg:logistic} logistic regression.
     \item \code{binary:logistic} logistic regression for binary classification. Output probability.
     \item \code{binary:logitraw} logistic regression for binary classification, output score before logistic transformation.
@@ -74,31 +75,31 @@ xgboost(data = NULL, label = NULL, missing = NA, weight = NULL,
 
 \item{watchlist}{named list of xgb.DMatrix datasets to use for evaluating model performance.
 Metrics specified in either \code{eval_metric} or \code{feval} will be computed for each
-of these datasets during each boosting iteration, and stored in the end as a field named 
-\code{evaluation_log} in the resulting object. When either \code{verbose>=1} or 
+of these datasets during each boosting iteration, and stored in the end as a field named
+\code{evaluation_log} in the resulting object. When either \code{verbose>=1} or
 \code{\link{cb.print.evaluation}} callback is engaged, the performance results are continuously
-printed out during the training. 
+printed out during the training.
 E.g., specifying \code{watchlist=list(validation1=mat1, validation2=mat2)} allows to track
 the performance of each round's model on mat1 and mat2.}
 
-\item{obj}{customized objective function. Returns gradient and second order 
+\item{obj}{customized objective function. Returns gradient and second order
 gradient with given prediction and dtrain.}
 
-\item{feval}{custimized evaluation function. Returns 
-\code{list(metric='metric-name', value='metric-value')} with given 
+\item{feval}{customized evaluation function. Returns
+\code{list(metric='metric-name', value='metric-value')} with given
 prediction and dtrain.}
 
 \item{verbose}{If 0, xgboost will stay silent. If 1, it will print information about performance.
 If 2, some additional information will be printed out.
-Note that setting \code{verbose > 0} automatically engages the 
+Note that setting \code{verbose > 0} automatically engages the
 \code{cb.print.evaluation(period=1)} callback function.}
 
 \item{print_every_n}{Print each n-th iteration evaluation messages when \code{verbose>0}.
-Default is 1 which means all messages are printed. This parameter is passed to the 
+Default is 1 which means all messages are printed. This parameter is passed to the
 \code{\link{cb.print.evaluation}} callback.}
 
-\item{early_stopping_rounds}{If \code{NULL}, the early stopping function is not triggered. 
-If set to an integer \code{k}, training with a validation set will stop if the performance 
+\item{early_stopping_rounds}{If \code{NULL}, the early stopping function is not triggered.
+If set to an integer \code{k}, training with a validation set will stop if the performance
 doesn't improve for \code{k} rounds.
 Setting this parameter engages the \code{\link{cb.early.stop}} callback.}
 
@@ -113,17 +114,17 @@ This parameter is passed to the \code{\link{cb.early.stop}} callback.}
 \item{save_name}{the name or path for periodically saved model file.}
 
 \item{xgb_model}{a previously built model to continue the training from.
-Could be either an object of class \code{xgb.Booster}, or its raw data, or the name of a 
+Could be either an object of class \code{xgb.Booster}, or its raw data, or the name of a
 file with a previously saved model.}
 
 \item{callbacks}{a list of callback functions to perform various task during boosting.
-See \code{\link{callbacks}}. Some of the callbacks are automatically created depending on the 
-parameters' values. User can provide either existing or their own callback methods in order 
+See \code{\link{callbacks}}. Some of the callbacks are automatically created depending on the
+parameters' values. User can provide either existing or their own callback methods in order
 to customize the training process.}
 
 \item{...}{other parameters to pass to \code{params}.}
 
-\item{label}{vector of response values. Should not be provided when data is 
+\item{label}{vector of response values. Should not be provided when data is
 a local data file name or an \code{xgb.DMatrix}.}
 
 \item{missing}{by default is set to NA, which means that NA values should be considered as 'missing'
@@ -138,23 +139,23 @@ An object of class \code{xgb.Booster} with the following elements:
   \item \code{handle} a handle (pointer) to the xgboost model in memory.
   \item \code{raw} a cached memory dump of the xgboost model saved as R's \code{raw} type.
   \item \code{niter} number of boosting iterations.
-  \item \code{evaluation_log} evaluation history storead as a \code{data.table} with the
+  \item \code{evaluation_log} evaluation history stored as a \code{data.table} with the
         first column corresponding to iteration number and the rest corresponding to evaluation
         metrics' values. It is created by the \code{\link{cb.evaluation.log}} callback.
   \item \code{call} a function call.
-  \item \code{params} parameters that were passed to the xgboost library. Note that it does not 
+  \item \code{params} parameters that were passed to the xgboost library. Note that it does not
         capture parameters changed by the \code{\link{cb.reset.parameters}} callback.
-  \item \code{callbacks} callback functions that were either automatically assigned or 
-        explicitely passed.
+  \item \code{callbacks} callback functions that were either automatically assigned or
+        explicitly passed.
   \item \code{best_iteration} iteration number with the best evaluation metric value
         (only available with early stopping).
-  \item \code{best_ntreelimit} the \code{ntreelimit} value corresponding to the best iteration, 
+  \item \code{best_ntreelimit} the \code{ntreelimit} value corresponding to the best iteration,
         which could further be used in \code{predict} method
         (only available with early stopping).
   \item \code{best_score} the best evaluation metric value during early stopping.
         (only available with early stopping).
   \item \code{feature_names} names of the training dataset features
-        (only when comun names were defined in training data).
+        (only when column names were defined in training data).
   \item \code{nfeatures} number of features in training data.
 }
 }
@@ -163,20 +164,20 @@ An object of class \code{xgb.Booster} with the following elements:
 The \code{xgboost} function is a simpler wrapper for \code{xgb.train}.
 }
 \details{
-These are the training functions for \code{xgboost}. 
+These are the training functions for \code{xgboost}.
 
-The \code{xgb.train} interface supports advanced features such as \code{watchlist}, 
-customized objective and evaluation metric functions, therefore it is more flexible 
+The \code{xgb.train} interface supports advanced features such as \code{watchlist},
+customized objective and evaluation metric functions, therefore it is more flexible
 than the \code{xgboost} interface.
 
-Parallelization is automatically enabled if \code{OpenMP} is present. 
+Parallelization is automatically enabled if \code{OpenMP} is present.
 Number of threads can also be manually specified via \code{nthread} parameter.
 
 The evaluation metric is chosen automatically by Xgboost (according to the objective)
 when the \code{eval_metric} parameter is not provided.
-User may set one or several \code{eval_metric} parameters. 
+User may set one or several \code{eval_metric} parameters.
 Note that when using a customized metric, only this single metric can be used.
-The folloiwing is the list of built-in metrics for which Xgboost provides optimized implementation:
+The following is the list of built-in metrics for which Xgboost provides optimized implementation:
   \itemize{
      \item \code{rmse} root mean square error. \url{http://en.wikipedia.org/wiki/Root_mean_square_error}
      \item \code{logloss} negative log-likelihood. \url{http://en.wikipedia.org/wiki/Log-likelihood}
@@ -208,7 +209,7 @@ dtest <- xgb.DMatrix(agaricus.test$data, label = agaricus.test$label)
 watchlist <- list(train = dtrain, eval = dtest)
 
 ## A simple xgb.train example:
-param <- list(max_depth = 2, eta = 1, silent = 1, nthread = 2, 
+param <- list(max_depth = 2, eta = 1, verbose = 0, nthread = 2,
               objective = "binary:logistic", eval_metric = "auc")
 bst <- xgb.train(param, dtrain, nrounds = 2, watchlist)
 
@@ -229,12 +230,12 @@ evalerror <- function(preds, dtrain) {
 
 # These functions could be used by passing them either:
 #  as 'objective' and 'eval_metric' parameters in the params list:
-param <- list(max_depth = 2, eta = 1, silent = 1, nthread = 2, 
+param <- list(max_depth = 2, eta = 1, verbose = 0, nthread = 2,
               objective = logregobj, eval_metric = evalerror)
 bst <- xgb.train(param, dtrain, nrounds = 2, watchlist)
 
 #  or through the ... arguments:
-param <- list(max_depth = 2, eta = 1, silent = 1, nthread = 2)
+param <- list(max_depth = 2, eta = 1, verbose = 0, nthread = 2)
 bst <- xgb.train(param, dtrain, nrounds = 2, watchlist,
                  objective = logregobj, eval_metric = evalerror)
 
@@ -244,7 +245,7 @@ bst <- xgb.train(param, dtrain, nrounds = 2, watchlist,
 
 
 ## An xgb.train example of using variable learning rates at each iteration:
-param <- list(max_depth = 2, eta = 1, silent = 1, nthread = 2,
+param <- list(max_depth = 2, eta = 1, verbose = 0, nthread = 2,
               objective = "binary:logistic", eval_metric = "auc")
 my_etas <- list(eta = c(0.5, 0.1))
 bst <- xgb.train(param, dtrain, nrounds = 2, watchlist,
@@ -255,8 +256,8 @@ bst <- xgb.train(param, dtrain, nrounds = 25, watchlist,
                  early_stopping_rounds = 3)
 
 ## An 'xgboost' interface example:
-bst <- xgboost(data = agaricus.train$data, label = agaricus.train$label, 
-               max_depth = 2, eta = 1, nthread = 2, nrounds = 2, 
+bst <- xgboost(data = agaricus.train$data, label = agaricus.train$label,
+               max_depth = 2, eta = 1, nthread = 2, nrounds = 2,
                objective = "binary:logistic")
 pred <- predict(bst, agaricus.test$data)
 

R-package/man/xgboost-deprecated.Rd

@@ -10,7 +10,7 @@ The deprecated parameters would be removed in the next release.
 \details{
 To see all the current deprecated and new parameters, check the \code{xgboost:::depr_par_lut} table.
 
-A deprecation warning is shown when any of the deprecated parameters is used in a call. 
-An additional warning is shown when there was a partial match to a deprecated parameter 
+A deprecation warning is shown when any of the deprecated parameters is used in a call.
+An additional warning is shown when there was a partial match to a deprecated parameter
 (as R is able to partially match parameter names).
 }

R-package/src/Makevars.in

@@ -12,13 +12,13 @@ XGB_RFLAGS = -DXGBOOST_STRICT_R_MODE=1 -DDMLC_LOG_BEFORE_THROW=0\
 
 # disable the use of thread_local for 32 bit windows:
 ifeq ($(R_OSTYPE)$(WIN),windows)
-    XGB_RFLAGS += -DDMLC_CXX11_THREAD_LOCAL=0 -msse2 -mfpmath=sse
+    XGB_RFLAGS += -DDMLC_CXX11_THREAD_LOCAL=0
 endif
 $(foreach v, $(XGB_RFLAGS), $(warning $(v)))
 
 PKG_CPPFLAGS=  -I$(PKGROOT)/include -I$(PKGROOT)/dmlc-core/include -I$(PKGROOT)/rabit/include -I$(PKGROOT) $(XGB_RFLAGS)
-PKG_CXXFLAGS= @OPENMP_CXXFLAGS@ $(SHLIB_PTHREAD_FLAGS)
-PKG_LIBS = @OPENMP_CXXFLAGS@ $(SHLIB_PTHREAD_FLAGS)
+PKG_CXXFLAGS= @OPENMP_CXXFLAGS@ -pthread
+PKG_LIBS = @OPENMP_CXXFLAGS@ -pthread
 OBJECTS= ./xgboost_R.o ./xgboost_custom.o ./xgboost_assert.o ./init.o\
          $(PKGROOT)/amalgamation/xgboost-all0.o $(PKGROOT)/amalgamation/dmlc-minimum0.o\
          $(PKGROOT)/rabit/src/engine_empty.o $(PKGROOT)/rabit/src/c_api.o

R-package/src/Makevars.win

@@ -24,13 +24,13 @@ XGB_RFLAGS = -DXGBOOST_STRICT_R_MODE=1 -DDMLC_LOG_BEFORE_THROW=0\
 
 # disable the use of thread_local for 32 bit windows:
 ifeq ($(R_OSTYPE)$(WIN),windows)
-    XGB_RFLAGS += -DDMLC_CXX11_THREAD_LOCAL=0 -msse2 -mfpmath=sse
+    XGB_RFLAGS += -DDMLC_CXX11_THREAD_LOCAL=0
 endif
 $(foreach v, $(XGB_RFLAGS), $(warning $(v)))
 
 PKG_CPPFLAGS=  -I$(PKGROOT)/include -I$(PKGROOT)/dmlc-core/include -I$(PKGROOT)/rabit/include -I$(PKGROOT) $(XGB_RFLAGS)
-PKG_CXXFLAGS= $(SHLIB_OPENMP_CFLAGS) $(SHLIB_PTHREAD_FLAGS)
-PKG_LIBS = $(SHLIB_OPENMP_CFLAGS) $(SHLIB_PTHREAD_FLAGS)
+PKG_CXXFLAGS= $(SHLIB_OPENMP_CXXFLAGS) $(SHLIB_PTHREAD_FLAGS)
+PKG_LIBS = $(SHLIB_OPENMP_CXXFLAGS) $(SHLIB_PTHREAD_FLAGS)
 OBJECTS= ./xgboost_R.o ./xgboost_custom.o ./xgboost_assert.o ./init.o\
          $(PKGROOT)/amalgamation/xgboost-all0.o $(PKGROOT)/amalgamation/dmlc-minimum0.o\
          $(PKGROOT)/rabit/src/engine_empty.o $(PKGROOT)/rabit/src/c_api.o

R-package/src/init.c

@@ -1,5 +1,5 @@
 /* Copyright (c) 2015 by Contributors
- * 
+ *
  * This file was initially generated using the following R command:
  * tools::package_native_routine_registration_skeleton('.', con = 'src/init.c', character_only = F)
  * and edited to conform to xgboost C linter requirements. For details, see
@@ -10,7 +10,7 @@
 #include <stdlib.h>
 #include <R_ext/Rdynload.h>
 
-/* FIXME: 
+/* FIXME:
 Check these declarations against the C/Fortran source code.
 */
 
@@ -70,7 +70,7 @@ static const R_CallMethodDef CallEntries[] = {
 
 #if defined(_WIN32)
 __declspec(dllexport)
-#endif
+#endif  // defined(_WIN32)
 void R_init_xgboost(DllInfo *dll) {
   R_registerRoutines(dll, NULL, CallEntries, NULL, NULL);
   R_useDynamicSymbols(dll, FALSE);

R-package/src/xgboost_R.h

@@ -1,6 +1,6 @@
 /*!
  * Copyright 2014 (c) by Contributors
- * \file xgboost_wrapper_R.h
+ * \file xgboost_R.h
  * \author Tianqi Chen
  * \brief R wrapper of xgboost
  */

R-package/src/xgboost_custom.cc

@@ -3,7 +3,7 @@
 // to change behavior of libxgboost
 
 #include <xgboost/logging.h>
-#include "src/common/random.h"
+#include "../../src/common/random.h"
 #include "./xgboost_R.h"
 
 // redirect the messages to R's console.
@@ -32,7 +32,10 @@ extern "C" {
 
 namespace xgboost {
 ConsoleLogger::~ConsoleLogger() {
-  dmlc::CustomLogMessage::Log(log_stream_.str());
+  if (cur_verbosity_ == LogVerbosity::kIgnore ||
+      cur_verbosity_ <= global_verbosity_) {
+    dmlc::CustomLogMessage::Log(log_stream_.str());
+  }
 }
 TrackerLogger::~TrackerLogger() {
   dmlc::CustomLogMessage::Log(log_stream_.str());
@@ -46,10 +49,11 @@ namespace common {
 bool CheckNAN(double v) {
   return ISNAN(v);
 }
+#if !defined(XGBOOST_USE_CUDA)
 double LogGamma(double v) {
   return lgammafn(v);
 }
-
+#endif  // !defined(XGBOOST_USE_CUDA)
 // customize random engine.
 void CustomGlobalRandomEngine::seed(CustomGlobalRandomEngine::result_type val) {
   // ignore the seed

R-package/tests/testthat/test_basic.R

@@ -182,7 +182,7 @@ test_that("xgb.cv works", {
   expect_is(cv, 'xgb.cv.synchronous')
   expect_false(is.null(cv$evaluation_log))
   expect_lt(cv$evaluation_log[, min(test_error_mean)], 0.03)
-  expect_lt(cv$evaluation_log[, min(test_error_std)], 0.004)
+  expect_lt(cv$evaluation_log[, min(test_error_std)], 0.008)
   expect_equal(cv$niter, 2)
   expect_false(is.null(cv$folds) && is.list(cv$folds))
   expect_length(cv$folds, 5)
@@ -223,3 +223,42 @@ test_that("train and predict with non-strict classes", {
   expect_error(pr <- predict(bst, train_dense), regexp = NA)
   expect_equal(pr0, pr)
 })
+
+test_that("max_delta_step works", {
+  dtrain <- xgb.DMatrix(agaricus.train$data, label = agaricus.train$label)
+  watchlist <- list(train = dtrain)
+  param <- list(objective = "binary:logistic", eval_metric="logloss", max_depth = 2, nthread = 2, eta = 0.5)
+  nrounds = 5
+  # model with no restriction on max_delta_step
+  bst1 <- xgb.train(param, dtrain, nrounds, watchlist, verbose = 1)
+  # model with restricted max_delta_step
+  bst2 <- xgb.train(param, dtrain, nrounds, watchlist, verbose = 1, max_delta_step = 1)
+  # the no-restriction model is expected to have consistently lower loss during the initial interations
+  expect_true(all(bst1$evaluation_log$train_logloss < bst2$evaluation_log$train_logloss))
+  expect_lt(mean(bst1$evaluation_log$train_logloss)/mean(bst2$evaluation_log$train_logloss), 0.8)
+})
+
+test_that("colsample_bytree works", {
+  # Randomly generate data matrix by sampling from uniform distribution [-1, 1]
+  set.seed(1)
+  train_x <- matrix(runif(1000, min = -1, max = 1), ncol = 100)
+  train_y <- as.numeric(rowSums(train_x) > 0)
+  test_x <- matrix(runif(1000, min = -1, max = 1), ncol = 100)
+  test_y <- as.numeric(rowSums(test_x) > 0)
+  colnames(train_x) <- paste0("Feature_", sprintf("%03d", 1:100))
+  colnames(test_x) <- paste0("Feature_", sprintf("%03d", 1:100))
+   dtrain <- xgb.DMatrix(train_x, label = train_y)
+  dtest <- xgb.DMatrix(test_x, label = test_y)
+  watchlist <- list(train = dtrain, eval = dtest)
+   # Use colsample_bytree = 0.01, so that roughly one out of 100 features is
+  # chosen for each tree
+  param <- list(max_depth = 2, eta = 0, silent = 1, nthread = 2,
+                colsample_bytree = 0.01, objective = "binary:logistic",
+                eval_metric = "auc")
+   set.seed(2)
+  bst <- xgb.train(param, dtrain, nrounds = 100, watchlist, verbose = 0)
+  xgb.importance(model = bst)
+  # If colsample_bytree works properly, a variety of features should be used
+  # in the 100 trees
+  expect_gte(nrow(xgb.importance(model = bst)), 30)
+})

R-package/tests/testthat/test_callbacks.R

@@ -236,7 +236,7 @@ test_that("early stopping using a specific metric works", {
   expect_equal(length(pred), 1611)
   logloss_pred <- sum(-ltest * log(pred) - (1 - ltest) * log(1 - pred)) / length(ltest)
   logloss_log <- bst$evaluation_log[bst$best_iteration, test_logloss]
-  expect_equal(logloss_log, logloss_pred, tolerance = 5e-6)
+  expect_equal(logloss_log, logloss_pred, tolerance = 1e-5)
 })
 
 test_that("early stopping xgb.cv works", {
@@ -282,7 +282,7 @@ test_that("prediction in xgb.cv works for gblinear too", {
 })
 
 test_that("prediction in early-stopping xgb.cv works", {
-  set.seed(1)
+  set.seed(11)
   expect_output(
     cv <- xgb.cv(param, dtrain, nfold = 5, eta = 0.1, nrounds = 20,
                  early_stopping_rounds = 5, maximize = FALSE, prediction = TRUE)

R-package/tests/testthat/test_custom_objective.R

@@ -31,7 +31,7 @@ num_round <- 2
 test_that("custom objective works", {
   bst <- xgb.train(param, dtrain, num_round, watchlist)
   expect_equal(class(bst), "xgb.Booster")
-  expect_equal(length(bst$raw), 1094)
+  expect_equal(length(bst$raw), 1100)
   expect_false(is.null(bst$evaluation_log))
   expect_false(is.null(bst$evaluation_log$eval_error))
   expect_lt(bst$evaluation_log[num_round, eval_error], 0.03)
@@ -45,7 +45,7 @@ test_that("custom objective in CV works", {
 })
 
 test_that("custom objective using DMatrix attr works", {
-  
+
   attr(dtrain, 'label') <- getinfo(dtrain, 'label')
 
   logregobjattr <- function(preds, dtrain) {
@@ -58,5 +58,5 @@ test_that("custom objective using DMatrix attr works", {
   param$objective = logregobjattr
   bst <- xgb.train(param, dtrain, num_round, watchlist)
   expect_equal(class(bst), "xgb.Booster")
-  expect_equal(length(bst$raw), 1094)
+  expect_equal(length(bst$raw), 1100)
 })

R-package/tests/testthat/test_dmatrix.R

@@ -10,12 +10,12 @@ test_label <- agaricus.test$label[1:100]
 test_that("xgb.DMatrix: basic construction", {
   # from sparse matrix
   dtest1 <- xgb.DMatrix(test_data, label=test_label)
-  
-  # from dense matrix 
+
+  # from dense matrix
   dtest2 <- xgb.DMatrix(as.matrix(test_data), label=test_label)
   expect_equal(getinfo(dtest1, 'label'), getinfo(dtest2, 'label'))
   expect_equal(dim(dtest1), dim(dtest2))
-  
+
   #from dense integer matrix
   int_data <- as.matrix(test_data)
   storage.mode(int_data) <- "integer"
@@ -33,7 +33,7 @@ test_that("xgb.DMatrix: saving, loading", {
   expect_output(dtest3 <- xgb.DMatrix(tmp_file, silent = TRUE), NA)
   unlink(tmp_file)
   expect_equal(getinfo(dtest1, 'label'), getinfo(dtest3, 'label'))
-  
+
   # from a libsvm text file
   tmp <- c("0 1:1 2:1","1 3:1","0 1:1")
   tmp_file <- 'tmp.libsvm'
@@ -49,7 +49,7 @@ test_that("xgb.DMatrix: getinfo & setinfo", {
   expect_true(setinfo(dtest, 'label', test_label))
   labels <- getinfo(dtest, 'label')
   expect_equal(test_label, getinfo(dtest, 'label'))
-  
+
   expect_true(length(getinfo(dtest, 'weight')) == 0)
   expect_true(length(getinfo(dtest, 'base_margin')) == 0)
 
@@ -57,10 +57,10 @@ test_that("xgb.DMatrix: getinfo & setinfo", {
   expect_true(setinfo(dtest, 'base_margin', test_label))
   expect_true(setinfo(dtest, 'group', c(50,50)))
   expect_error(setinfo(dtest, 'group', test_label))
-  
+
   # providing character values will give a warning
   expect_warning( setinfo(dtest, 'weight', rep('a', nrow(test_data))) )
-  
+
   # any other label should error
   expect_error(setinfo(dtest, 'asdf', test_label))
 })
@@ -71,7 +71,7 @@ test_that("xgb.DMatrix: slice, dim", {
   dsub1 <- slice(dtest, 1:42)
   expect_equal(nrow(dsub1), 42)
   expect_equal(ncol(dsub1), ncol(test_data))
-  
+
   dsub2 <- dtest[1:42,]
   expect_equal(dim(dtest), dim(test_data))
   expect_equal(getinfo(dsub1, 'label'), getinfo(dsub2, 'label'))

R-package/tests/testthat/test_gc_safety.R

@@ -9,7 +9,7 @@ test_that("train and prediction when gctorture is on", {
   test <- agaricus.test
   gctorture(TRUE)
   bst <- xgboost(data = train$data, label = train$label, max.depth = 2,
-  eta = 1, nthread = 2, nround = 2, objective = "binary:logistic")
+  eta = 1, nthread = 2, nrounds = 2, objective = "binary:logistic")
   pred <- predict(bst, test$data)
   gctorture(FALSE)
 })

R-package/tests/testthat/test_helpers.R

@@ -7,6 +7,9 @@ require(vcd, quietly = TRUE)
 
 float_tolerance = 5e-6
 
+# disable some tests for Win32
+win32_flag = .Platform$OS.type == "windows" && .Machine$sizeof.pointer != 8
+
 set.seed(1982)
 data(Arthritis)
 df <- data.table(Arthritis, keep.rownames = F)
@@ -41,7 +44,8 @@ mbst.GLM <- xgboost(data = as.matrix(iris[, -5]), label = mlabel, verbose = 0,
 
 
 test_that("xgb.dump works", {
-  expect_length(xgb.dump(bst.Tree), 200)
+  if (!win32_flag)
+    expect_length(xgb.dump(bst.Tree), 200)
   dump_file = file.path(tempdir(), 'xgb.model.dump')
   expect_true(xgb.dump(bst.Tree, dump_file, with_stats = T))
   expect_true(file.exists(dump_file))
@@ -50,7 +54,8 @@ test_that("xgb.dump works", {
   # JSON format
   dmp <- xgb.dump(bst.Tree, dump_format = "json")
   expect_length(dmp, 1)
-  expect_length(grep('nodeid', strsplit(dmp, '\n')[[1]]), 188)
+  if (!win32_flag)
+    expect_length(grep('nodeid', strsplit(dmp, '\n')[[1]]), 188)
 })
 
 test_that("xgb.dump works for gblinear", {
@@ -158,6 +163,7 @@ test_that("xgb-attribute functionality", {
   # serializing:
   xgb.save(bst.Tree, 'xgb.model')
   bst <- xgb.load('xgb.model')
+  if (file.exists('xgb.model')) file.remove('xgb.model')
   expect_equal(xgb.attr(bst, "my_attr"), val)
   expect_equal(xgb.attributes(bst), list.ch)
   # deletion:
@@ -194,10 +200,12 @@ if (grepl('Windows', Sys.info()[['sysname']]) ||
 test_that("xgb.Booster serializing as R object works", {
   saveRDS(bst.Tree, 'xgb.model.rds')
   bst <- readRDS('xgb.model.rds')
+  if (file.exists('xgb.model.rds')) file.remove('xgb.model.rds')
   dtrain <- xgb.DMatrix(sparse_matrix, label = label)
   expect_equal(predict(bst.Tree, dtrain), predict(bst, dtrain), tolerance = float_tolerance)
   expect_equal(xgb.dump(bst.Tree), xgb.dump(bst))
   xgb.save(bst, 'xgb.model')
+  if (file.exists('xgb.model')) file.remove('xgb.model')
   nil_ptr <- new("externalptr")
   class(nil_ptr) <- "xgb.Booster.handle"
   expect_true(identical(bst$handle, nil_ptr))
@@ -210,7 +218,8 @@ test_that("xgb.model.dt.tree works with and without feature names", {
   names.dt.trees <- c("Tree", "Node", "ID", "Feature", "Split", "Yes", "No", "Missing", "Quality", "Cover")
   dt.tree <- xgb.model.dt.tree(feature_names = feature.names, model = bst.Tree)
   expect_equal(names.dt.trees, names(dt.tree))
-  expect_equal(dim(dt.tree), c(188, 10))
+  if (!win32_flag)
+    expect_equal(dim(dt.tree), c(188, 10))
   expect_output(str(dt.tree), 'Feature.*\\"Age\\"')
 
   dt.tree.0 <- xgb.model.dt.tree(model = bst.Tree)
@@ -236,7 +245,8 @@ test_that("xgb.model.dt.tree throws error for gblinear", {
 
 test_that("xgb.importance works with and without feature names", {
   importance.Tree <- xgb.importance(feature_names = feature.names, model = bst.Tree)
-  expect_equal(dim(importance.Tree), c(7, 4))
+  if (!win32_flag)
+    expect_equal(dim(importance.Tree), c(7, 4))
   expect_equal(colnames(importance.Tree), c("Feature", "Gain", "Cover", "Frequency"))
   expect_output(str(importance.Tree), 'Feature.*\\"Age\\"')
 

R-package/tests/testthat/test_interaction_constraints.R

@@ -0,0 +1,38 @@
+require(xgboost)
+
+context("interaction constraints")
+
+set.seed(1024)
+x1 <- rnorm(1000, 1)
+x2 <- rnorm(1000, 1)
+x3 <- sample(c(1,2,3), size=1000, replace=TRUE)
+y <- x1 + x2 + x3 + x1*x2*x3 + rnorm(1000, 0.001) + 3*sin(x1)
+train <- matrix(c(x1,x2,x3), ncol = 3)
+
+test_that("interaction constraints for regression", {
+  # Fit a model that only allows interaction between x1 and x2
+  bst <- xgboost(data = train, label = y, max_depth = 3,
+                 eta = 0.1, nthread = 2, nrounds = 100, verbose = 0,
+                 interaction_constraints = list(c(0,1)))
+  
+  # Set all observations to have the same x3 values then increment
+  #  by the same amount
+	preds <- lapply(c(1,2,3), function(x){
+		tmat <- matrix(c(x1,x2,rep(x,1000)), ncol=3)
+		return(predict(bst, tmat))
+	})
+
+  # Check incrementing x3 has the same effect on all observations
+  #   since x3 is constrained to be independent of x1 and x2
+  #   and all observations start off from the same x3 value
+	diff1 <- preds[[2]] - preds[[1]]
+	test1 <- all(abs(diff1 - diff1[1]) < 1e-4)
+	
+	diff2 <- preds[[3]] - preds[[2]]
+	test2 <- all(abs(diff2 - diff2[1]) < 1e-4)
+	
+  expect_true({
+    test1 & test2
+  }, "Interaction Contraint Satisfied")
+  
+})

R-package/tests/testthat/test_interactions.R

@@ -0,0 +1,141 @@
+context('Test prediction of feature interactions')
+
+require(xgboost)
+require(magrittr)
+
+set.seed(123)
+
+test_that("predict feature interactions works", {
+  # simulate some binary data and a linear outcome with an interaction term
+  N <- 1000
+  P <- 5
+  X <- matrix(rbinom(N * P, 1, 0.5), ncol=P, dimnames = list(NULL, letters[1:P]))
+  # center the data (as contributions are computed WRT feature means)
+  X <- scale(X, scale=FALSE)
+
+  # outcome without any interactions, without any noise:
+  f <- function(x) 2 * x[, 1] - 3 * x[, 2]
+  # outcome with interactions, without noise:
+  f_int <- function(x) f(x) + 2 * x[, 2] * x[, 3]
+  # outcome with interactions, with noise:
+  #f_int_noise <- function(x) f_int(x) + rnorm(N, 0, 0.3)
+
+  y <- f_int(X)
+
+  dm <- xgb.DMatrix(X, label = y)
+  param <- list(eta=0.1, max_depth=4, base_score=mean(y), lambda=0, nthread=2)
+  b <- xgb.train(param, dm, 100)
+  
+  pred = predict(b, dm, outputmargin=TRUE)
+
+  # SHAP contributions:
+  cont <- predict(b, dm, predcontrib=TRUE)
+  expect_equal(dim(cont), c(N, P+1))
+  # make sure for each row they add up to marginal predictions
+  max(abs(rowSums(cont) - pred)) %>% expect_lt(0.001)
+  # Hand-construct the 'ground truth' feature contributions:
+  gt_cont <- cbind(
+      2. * X[, 1],
+     -3. * X[, 2] + 1. * X[, 2] * X[, 3], # attribute a HALF of the interaction term to feature #2
+      1. * X[, 2] * X[, 3]               # and another HALF of the interaction term to feature #3
+     )
+  gt_cont <- cbind(gt_cont, matrix(0, nrow=N, ncol=P + 1 - 3))
+  # These should be relatively close:
+  expect_lt(max(abs(cont - gt_cont)), 0.05)
+
+
+  # SHAP interaction contributions:
+  intr <- predict(b, dm, predinteraction=TRUE)
+  expect_equal(dim(intr), c(N, P+1, P+1))
+  # check assigned colnames
+  cn <- c(letters[1:P], "BIAS")
+  expect_equal(dimnames(intr), list(NULL, cn, cn))
+
+  # check the symmetry
+  max(abs(aperm(intr, c(1,3,2)) - intr)) %>% expect_lt(0.00001)
+
+  # sums WRT columns must be close to feature contributions
+  max(abs(apply(intr, c(1,2), sum) - cont)) %>% expect_lt(0.00001)
+
+  # diagonal terms for features 3,4,5 must be close to zero
+  Reduce(max, sapply(3:P, function(i) max(abs(intr[, i, i])))) %>% expect_lt(0.05)
+
+  # BIAS must have no interactions
+  max(abs(intr[, 1:P, P+1])) %>% expect_lt(0.00001)
+
+  # interactions other than 2 x 3 must be close to zero
+  intr23 <- intr
+  intr23[,2,3] <- 0
+  Reduce(max, sapply(1:P, function(i) max(abs(intr23[, i, (i+1):(P+1)])))) %>% expect_lt(0.05)
+
+  # Construct the 'ground truth' contributions of interactions directly from the linear terms:
+  gt_intr <- array(0, c(N, P+1, P+1))
+  gt_intr[,2,3] <- 1. * X[, 2] * X[, 3] # attribute a HALF of the interaction term to each symmetric element
+  gt_intr[,3,2] <- gt_intr[, 2, 3]
+  # merge-in the diagonal based on 'ground truth' feature contributions
+  intr_diag = gt_cont - apply(gt_intr, c(1,2), sum)
+  for(j in seq_len(P)) {
+    gt_intr[,j,j] = intr_diag[,j]
+  }
+  # These should be relatively close:
+  expect_lt(max(abs(intr - gt_intr)), 0.1)
+})
+
+test_that("SHAP contribution values are not NAN", {
+  d <- data.frame(
+    x1 = c(-2.3, 1.4, 5.9, 2, 2.5, 0.3, -3.6, -0.2, 0.5, -2.8, -4.6, 3.3, -1.2,
+           -1.1, -2.3, 0.4, -1.5, -0.2, -1, 3.7),
+    x2 = c(291.179171, 269.198331, 289.942097, 283.191669, 269.673332,
+           294.158346, 287.255835, 291.530838, 285.899586, 269.290833,
+           268.649586, 291.530841, 280.074593, 269.484168, 293.94042,
+           294.327506, 296.20709, 295.441669, 283.16792, 270.227085),
+    y = c(9, 15, 5.7, 9.2, 22.4, 5, 9, 3.2, 7.2, 13.1, 7.8, 16.9, 6.5, 22.1,
+          5.3, 10.4, 11.1, 13.9, 11, 20.5),
+    fold = c(2, 2, 2, 1, 2, 2, 1, 2, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2))
+
+  ivs <- c("x1", "x2")
+
+  fit <- xgboost(
+    verbose = 0,
+    params = list(
+      objective = "reg:squarederror",
+      eval_metric = "rmse"),
+    data = as.matrix(subset(d, fold == 2)[, ivs]),
+    label = subset(d, fold == 2)$y,
+    nthread = 1,
+    nrounds = 3)
+
+  shaps <- as.data.frame(predict(fit,
+    newdata = as.matrix(subset(d, fold == 1)[, ivs]),
+    predcontrib = T))
+  result <- cbind(shaps, sum = rowSums(shaps), pred = predict(fit,
+      newdata = as.matrix(subset(d, fold == 1)[, ivs])))
+
+  expect_true(identical(TRUE, all.equal(result$sum, result$pred, tol = 1e-6)))
+})
+
+
+test_that("multiclass feature interactions work", {
+  dm <- xgb.DMatrix(as.matrix(iris[,-5]), label=as.numeric(iris$Species)-1)
+  param <- list(eta=0.1, max_depth=4, objective='multi:softprob', num_class=3)
+  b <- xgb.train(param, dm, 40)
+  pred = predict(b, dm, outputmargin=TRUE) %>% array(c(3, 150)) %>% t
+
+  # SHAP contributions:
+  cont <- predict(b, dm, predcontrib=TRUE)
+  expect_length(cont, 3)
+  # rewrap them as a 3d array
+  cont <- unlist(cont) %>% array(c(150, 5, 3))
+  # make sure for each row they add up to marginal predictions
+  max(abs(apply(cont, c(1,3), sum) - pred)) %>% expect_lt(0.001)
+
+  # SHAP interaction contributions:
+  intr <- predict(b, dm, predinteraction=TRUE)
+  expect_length(intr, 3)
+  # rewrap them as a 4d array
+  intr <- unlist(intr) %>% array(c(150, 5, 5, 3)) %>% aperm(c(4, 1, 2, 3)) # [grp, row, col, col]
+  # check the symmetry
+  max(abs(aperm(intr, c(1,2,4,3)) - intr)) %>% expect_lt(0.00001)
+  # sums WRT columns must be close to feature contributions
+  max(abs(apply(intr, c(1,2,3), sum) - aperm(cont, c(3,1,2)))) %>% expect_lt(0.00001)
+})

R-package/tests/testthat/test_update.R

@@ -7,6 +7,10 @@ data(agaricus.test, package = 'xgboost')
 dtrain <- xgb.DMatrix(agaricus.train$data, label = agaricus.train$label)
 dtest <- xgb.DMatrix(agaricus.test$data, label = agaricus.test$label)
 
+# Disable flaky tests for 32-bit Windows.
+# See https://github.com/dmlc/xgboost/issues/3720
+win32_flag = .Platform$OS.type == "windows" && .Machine$sizeof.pointer != 8
+
 test_that("updating the model works", {
   watchlist = list(train = dtrain, test = dtest)
 
@@ -29,7 +33,9 @@ test_that("updating the model works", {
   tr1r <- xgb.model.dt.tree(model = bst1r)
   # all should be the same when no subsampling
   expect_equal(bst1$evaluation_log, bst1r$evaluation_log)
-  expect_equal(tr1, tr1r, tolerance = 0.00001, check.attributes = FALSE)
+  if (!win32_flag) {
+    expect_equal(tr1, tr1r, tolerance = 0.00001, check.attributes = FALSE)
+  }
 
   # the same boosting with subsampling with an extra 'refresh' updater:
   p2r <- modifyList(p2, list(updater = 'grow_colmaker,prune,refresh', refresh_leaf = FALSE))
@@ -38,7 +44,9 @@ test_that("updating the model works", {
   tr2r <- xgb.model.dt.tree(model = bst2r)
   # should be the same evaluation but different gains and larger cover
   expect_equal(bst2$evaluation_log, bst2r$evaluation_log)
-  expect_equal(tr2[Feature == 'Leaf']$Quality, tr2r[Feature == 'Leaf']$Quality)
+  if (!win32_flag) {
+    expect_equal(tr2[Feature == 'Leaf']$Quality, tr2r[Feature == 'Leaf']$Quality)
+  }
   expect_gt(sum(abs(tr2[Feature != 'Leaf']$Quality - tr2r[Feature != 'Leaf']$Quality)), 100)
   expect_gt(sum(tr2r$Cover) / sum(tr2$Cover), 1.5)
 
@@ -61,7 +69,9 @@ test_that("updating the model works", {
   expect_gt(sum(tr2u$Cover) / sum(tr2$Cover), 1.5)
   # the results should be the same as for the model with an extra 'refresh' updater
   expect_equal(bst2r$evaluation_log, bst2u$evaluation_log)
-  expect_equal(tr2r, tr2u, tolerance = 0.00001, check.attributes = FALSE)
+  if (!win32_flag) {
+    expect_equal(tr2r, tr2u, tolerance = 0.00001, check.attributes = FALSE)
+  }
   
   # process type 'update' for no-subsampling model, refreshing only the tree stats from TEST data:
   p1ut <- modifyList(p1, list(process_type = 'update', updater = 'refresh', refresh_leaf = FALSE))

R-package/vignettes/discoverYourData.Rmd

@@ -138,7 +138,7 @@ levels(df[,Treatment])
 
 Next step, we will transform the categorical data to dummy variables.
 Several encoding methods exist, e.g., [one-hot encoding](http://en.wikipedia.org/wiki/One-hot) is a common approach.
-We will use the [dummy contrast coding](http://www.ats.ucla.edu/stat/r/library/contrast_coding.htm#dummy) which is popular because it producess "full rank" encoding (also see [this blog post by Max Kuhn](http://appliedpredictivemodeling.com/blog/2013/10/23/the-basics-of-encoding-categorical-data-for-predictive-models)).
+We will use the [dummy contrast coding](http://www.ats.ucla.edu/stat/r/library/contrast_coding.htm#dummy) which is popular because it produces "full rank" encoding (also see [this blog post by Max Kuhn](http://appliedpredictivemodeling.com/blog/2013/10/23/the-basics-of-encoding-categorical-data-for-predictive-models)).
 
 The purpose is to transform each value of each *categorical* feature into a *binary* feature `{0, 1}`.
 
@@ -268,7 +268,7 @@ c2 <- chisq.test(df$Age, output_vector)
 print(c2)
 ```
 
-Pearson correlation between Age and illness disapearing is **`r round(c2$statistic, 2 )`**.
+Pearson correlation between Age and illness disappearing is **`r round(c2$statistic, 2 )`**.
 
 ```{r, warning=FALSE, message=FALSE}
 c2 <- chisq.test(df$AgeDiscret, output_vector)

R-package/vignettes/xgboostPresentation.Rmd

@@ -313,7 +313,7 @@ Until now, all the learnings we have performed were based on boosting trees. **X
 bst <- xgb.train(data=dtrain, booster = "gblinear", max_depth=2, nthread = 2, nrounds=2, watchlist=watchlist, eval_metric = "error", eval_metric = "logloss", objective = "binary:logistic")
 ```
 
-In this specific case, *linear boosting* gets sligtly better performance metrics than decision trees based algorithm.
+In this specific case, *linear boosting* gets slightly better performance metrics than decision trees based algorithm.
 
 In simple cases, it will happen because there is nothing better than a linear algorithm to catch a linear link. However, decision trees are much better to catch a non linear link between predictors and outcome. Because there is no silver bullet, we advise you to check both algorithms with your own datasets to have an idea of what to use.
 

R-package/vignettes/xgboostfromJSON.Rmd

@@ -0,0 +1,189 @@
+---
+title: "XGBoost from JSON"
+output:
+  rmarkdown::html_vignette:
+    number_sections: yes
+    toc: yes
+author: Roland Stevenson
+vignette: >
+  %\VignetteIndexEntry{XGBoost from JSON}
+  %\VignetteEngine{knitr::rmarkdown}
+  \usepackage[utf8]{inputenc}
+---
+
+XGBoost from JSON
+=================
+
+## Introduction
+
+The purpose of this Vignette is to show you how to correctly load and work with an **Xgboost** model that has been dumped to JSON.  **Xgboost** internally converts all data to [32-bit floats](https://en.wikipedia.org/wiki/Single-precision_floating-point_format), and the values dumped to JSON are decimal representations of these values.  When working with a model that has been parsed from a JSON file, care must be taken to correctly treat:
+
+- the input data, which should be converted to 32-bit floats
+- any 32-bit floats that were stored in JSON as decimal representations
+- any calculations must be done with 32-bit mathematical operators
+
+## Setup
+
+For the purpose of this tutorial we will load the xgboost, jsonlite, and float packages.  We'll also set `digits=22` in our options in case we want to inspect many digits of our results.
+
+```{r}
+require(xgboost)
+require(jsonlite)
+require(float)
+options(digits=22)
+```
+
+We will create a toy binary logistic model based on the example first provided [here](https://github.com/dmlc/xgboost/issues/3960), so that we can easily understand the structure of the dumped JSON model object.  This will allow us to understand where discrepancies can occur and how they should be handled.
+
+```{r}
+dates <- c(20180130, 20180130, 20180130,
+           20180130, 20180130, 20180130,
+           20180131, 20180131, 20180131,
+           20180131, 20180131, 20180131,
+           20180131, 20180131, 20180131,
+           20180134, 20180134, 20180134)
+
+labels <- c(1, 1, 1,
+            1, 1, 1,
+            0, 0, 0,
+            0, 0, 0,
+            0, 0, 0,
+            0, 0, 0)
+
+data <- data.frame(dates = dates, labels=labels)
+
+bst <- xgboost(
+  data = as.matrix(data$dates), 
+  label = labels,
+  nthread = 2,
+  nrounds = 1,
+  objective = "binary:logistic",
+  missing = NA,
+  max_depth = 1
+)
+```
+
+## Comparing results
+We will now dump the model to JSON and attempt to illustrate a variety of issues that can arise, and how to properly deal with them.
+
+First let's dump the model to JSON:
+
+```{r}
+bst_json <- xgb.dump(bst, with_stats = FALSE, dump_format='json')
+bst_from_json <- fromJSON(bst_json, simplifyDataFrame = FALSE)
+node <- bst_from_json[[1]]
+cat(bst_json)
+```
+
+The tree JSON shown by the above code-chunk tells us that if the data is less than 20180132, the tree will output the value in the first leaf.  Otherwise it will output the value in the second leaf.  Let's try to reproduce this manually with the data we have and confirm that it matches the model predictions we've already calculated.
+
+```{r}
+bst_preds_logodds <- predict(bst,as.matrix(data$dates), outputmargin = TRUE)
+
+# calculate the logodds values using the JSON representation
+bst_from_json_logodds <- ifelse(data$dates<node$split_condition,
+                                node$children[[1]]$leaf,
+                                node$children[[2]]$leaf)
+
+bst_preds_logodds
+bst_from_json_logodds
+
+# test that values are equal
+bst_preds_logodds == bst_from_json_logodds
+
+```
+None are equal.  What happened?
+
+At this stage two things happened:
+
+- input data was not converted to 32-bit floats
+- the JSON variables were not converted to 32-bit floats
+
+### Lesson 1: All data is 32-bit floats
+
+> When working with imported JSON, all data must be converted to 32-bit floats
+
+To explain this, let's repeat the comparison and round to two decimals:
+
+```{r}
+round(bst_preds_logodds,2) == round(bst_from_json_logodds,2)
+```
+
+If we round to two decimals, we see that only the elements related to data values of `20180131` don't agree.  If we convert the data to floats, they agree:
+
+```{r}
+# now convert the dates to floats first
+bst_from_json_logodds <- ifelse(fl(data$dates)<node$split_condition,
+                                node$children[[1]]$leaf,
+                                node$children[[2]]$leaf)
+
+# test that values are equal
+round(bst_preds_logodds,2) == round(bst_from_json_logodds,2)
+```
+
+What's the lesson?  If we are going to work with an imported JSON model, any data must be converted to floats first.  In this case, since '20180131' cannot be represented as a 32-bit float, it is rounded up to 20180132, as shown here:
+
+```{r}
+fl(20180131)
+```
+
+
+### Lesson 2: JSON parameters are 32-bit floats
+
+> All JSON parameters stored as floats must be converted to floats.
+
+Let's now say we do care about numbers past the first two decimals.
+
+```{r}
+# test that values are equal
+bst_preds_logodds == bst_from_json_logodds
+```
+
+None are exactly equal.  What happened?  Although we've converted the data to 32-bit floats, we also need to convert the JSON parameters to 32-bit floats.  Let's do this:
+
+```{r}
+# now convert the dates to floats first
+bst_from_json_logodds <- ifelse(fl(data$dates)<fl(node$split_condition),
+                                as.numeric(fl(node$children[[1]]$leaf)),
+                                as.numeric(fl(node$children[[2]]$leaf)))
+
+# test that values are equal
+bst_preds_logodds == bst_from_json_logodds
+```
+All equal.  What's the lesson?  If we are going to work with an imported JSON model, any JSON parameters that were stored as floats must also be converted to floats first.
+
+### Lesson 3: Use 32-bit math
+
+> Always use 32-bit numbers and operators
+
+We were able to get the log-odds to agree, so now let's manually calculate the sigmoid of the log-odds.  This should agree with the xgboost predictions.
+
+
+```{r}
+bst_preds <- predict(bst,as.matrix(data$dates))
+
+# calculate the predictions casting doubles to floats
+bst_from_json_preds <- ifelse(fl(data$dates)<fl(node$split_condition),
+                              as.numeric(1/(1+exp(-1*fl(node$children[[1]]$leaf)))),
+                              as.numeric(1/(1+exp(-1*fl(node$children[[2]]$leaf))))
+)
+
+# test that values are equal
+bst_preds == bst_from_json_preds
+```
+
+None are exactly equal again.  What is going on here?  Well, since we are using the value `1` in the calcuations, we have introduced a double into the calculation.  Because of this, all float values are promoted to 64-bit doubles and the 64-bit version of the exponentiation operator `exp` is also used.  On the other hand, xgboost uses the 32-bit version of the exponentation operator in its [sigmoid function](https://github.com/dmlc/xgboost/blob/54980b8959680a0da06a3fc0ec776e47c8cbb0a1/src/common/math.h#L25-L27).
+
+How do we fix this?  We have to ensure we use the correct datatypes everywhere and the correct operators.  If we use only floats, the float library that we have loaded will ensure the 32-bit float exponention operator is applied. 
+```{r}
+# calculate the predictions casting doubles to floats
+bst_from_json_preds <- ifelse(fl(data$dates)<fl(node$split_condition),
+                              as.numeric(fl(1)/(fl(1)+exp(fl(-1)*fl(node$children[[1]]$leaf)))),
+                              as.numeric(fl(1)/(fl(1)+exp(fl(-1)*fl(node$children[[2]]$leaf))))
+)
+
+# test that values are equal
+bst_preds == bst_from_json_preds
+```
+
+All equal.  What's the lesson?  We have to ensure that all calculations are done with 32-bit floating point operators if we want to reproduce the results that we see with xgboost.

README.md

@@ -1,6 +1,7 @@
 <img src=https://raw.githubusercontent.com/dmlc/dmlc.github.io/master/img/logo-m/xgboost.png width=135/>  eXtreme Gradient Boosting
 ===========
-[![Build Status](https://travis-ci.org/dmlc/xgboost.svg?branch=master)](https://travis-ci.org/dmlc/xgboost)
+[![Build Status](https://xgboost-ci.net/job/xgboost/job/master/badge/icon?style=plastic)](https://xgboost-ci.net/blue/organizations/jenkins/xgboost/activity)
+[![Build Status](https://img.shields.io/travis/dmlc/xgboost.svg?label=build&logo=travis&branch=master)](https://travis-ci.org/dmlc/xgboost)
 [![Build Status](https://ci.appveyor.com/api/projects/status/5ypa8vaed6kpmli8?svg=true)](https://ci.appveyor.com/project/tqchen/xgboost)
 [![Documentation Status](https://readthedocs.org/projects/xgboost/badge/?version=latest)](https://xgboost.readthedocs.org)
 [![GitHub license](http://dmlc.github.io/img/apache2.svg)](./LICENSE)
@@ -31,3 +32,35 @@ Reference
 ---------
 - Tianqi Chen and Carlos Guestrin. [XGBoost: A Scalable Tree Boosting System](http://arxiv.org/abs/1603.02754). In 22nd SIGKDD Conference on Knowledge Discovery and Data Mining, 2016
 - XGBoost originates from research project at University of Washington.
+
+Sponsors
+--------
+Become a sponsor and get a logo here. See details at [Sponsoring the XGBoost Project](https://xgboost.ai/sponsors). The funds are used to defray the cost of continuous integration and testing infrastructure (https://xgboost-ci.net).
+
+## Open Source Collective sponsors
+[![Backers on Open Collective](https://opencollective.com/xgboost/backers/badge.svg)](#backers) [![Sponsors on Open Collective](https://opencollective.com/xgboost/sponsors/badge.svg)](#sponsors) 
+
+### Sponsors
+[[Become a sponsor](https://opencollective.com/xgboost#sponsor)]
+
+<!--<a href="https://opencollective.com/xgboost/sponsor/0/website" target="_blank"><img src="https://opencollective.com/xgboost/sponsor/0/avatar.svg"></a>-->
+<a href="https://www.nvidia.com/en-us/" target="_blank"><img src="https://raw.githubusercontent.com/xgboost-ai/xgboost-ai.github.io/master/images/sponsors/nvidia.jpg" alt="NVIDIA" width="72" height="72"></a>
+<a href="https://opencollective.com/xgboost/sponsor/1/website" target="_blank"><img src="https://opencollective.com/xgboost/sponsor/1/avatar.svg"></a>
+<a href="https://opencollective.com/xgboost/sponsor/2/website" target="_blank"><img src="https://opencollective.com/xgboost/sponsor/2/avatar.svg"></a>
+<a href="https://opencollective.com/xgboost/sponsor/3/website" target="_blank"><img src="https://opencollective.com/xgboost/sponsor/3/avatar.svg"></a>
+<a href="https://opencollective.com/xgboost/sponsor/4/website" target="_blank"><img src="https://opencollective.com/xgboost/sponsor/4/avatar.svg"></a>
+<a href="https://opencollective.com/xgboost/sponsor/5/website" target="_blank"><img src="https://opencollective.com/xgboost/sponsor/5/avatar.svg"></a>
+<a href="https://opencollective.com/xgboost/sponsor/6/website" target="_blank"><img src="https://opencollective.com/xgboost/sponsor/6/avatar.svg"></a>
+<a href="https://opencollective.com/xgboost/sponsor/7/website" target="_blank"><img src="https://opencollective.com/xgboost/sponsor/7/avatar.svg"></a>
+<a href="https://opencollective.com/xgboost/sponsor/8/website" target="_blank"><img src="https://opencollective.com/xgboost/sponsor/8/avatar.svg"></a>
+<a href="https://opencollective.com/xgboost/sponsor/9/website" target="_blank"><img src="https://opencollective.com/xgboost/sponsor/9/avatar.svg"></a>
+
+### Backers
+[[Become a backer](https://opencollective.com/xgboost#backer)]
+
+<a href="https://opencollective.com/xgboost#backers" target="_blank"><img src="https://opencollective.com/xgboost/backers.svg?width=890"></a>
+
+## Other sponsors
+The sponsors in this list are donating cloud hours in lieu of cash donation.
+
+<a href="https://aws.amazon.com/" target="_blank"><img src="https://raw.githubusercontent.com/xgboost-ai/xgboost-ai.github.io/master/images/sponsors/aws.png" alt="Amazon Web Services" width="72" height="72"></a>

amalgamation/xgboost-all0.cc

@@ -48,7 +48,7 @@
 #include "../src/tree/tree_model.cc"
 #include "../src/tree/tree_updater.cc"
 #include "../src/tree/updater_colmaker.cc"
-#include "../src/tree/updater_fast_hist.cc"
+#include "../src/tree/updater_quantile_hist.cc"
 #include "../src/tree/updater_prune.cc"
 #include "../src/tree/updater_refresh.cc"
 #include "../src/tree/updater_sync.cc"

appveyor.yml

@@ -36,26 +36,32 @@ install:
     - set PATH=C:\msys64\mingw64\bin;C:\msys64\usr\bin;%PATH%
     - gcc -v
     - ls -l C:\
-    # Miniconda2
-    - set PATH=;C:\Miniconda-x64;C:\Miniconda-x64\Scripts;%PATH%
+    # Miniconda3
+    - call C:\Miniconda3-x64\Scripts\activate.bat
+    - conda info
     - where python
     - python --version
     # do python build for mingw and one of the msvc jobs
     - set DO_PYTHON=off
     - if /i "%target%" == "mingw" set DO_PYTHON=on
     - if /i "%target%_%ver%_%configuration%" == "msvc_2015_Release" set DO_PYTHON=on
-    - if /i "%DO_PYTHON%" == "on" conda install -y numpy scipy pandas matplotlib nose scikit-learn graphviz python-graphviz
+    - if /i "%DO_PYTHON%" == "on" (
+        conda config --set always_yes true &&
+        conda update -q conda &&
+        conda install -y numpy scipy pandas matplotlib pytest scikit-learn graphviz python-graphviz
+      )
+    - set PATH=C:\Miniconda3-x64\Library\bin\graphviz;%PATH%
     # R: based on https://github.com/krlmlr/r-appveyor
     - ps: |
         if($env:target -eq 'rmingw' -or $env:target -eq 'rmsvc') {
           #$ErrorActionPreference = "Stop"
-          Invoke-WebRequest http://raw.github.com/krlmlr/r-appveyor/master/scripts/appveyor-tool.ps1 -OutFile "$Env:TEMP\appveyor-tool.ps1"
+          Invoke-WebRequest https://raw.githubusercontent.com/krlmlr/r-appveyor/master/scripts/appveyor-tool.ps1 -OutFile "$Env:TEMP\appveyor-tool.ps1"
           Import-Module "$Env:TEMP\appveyor-tool.ps1"
           Bootstrap
-          $DEPS = "c('data.table','magrittr','stringi','ggplot2','DiagrammeR','Ckmeans.1d.dp','vcd','testthat','lintr','knitr','rmarkdown')"
-          cmd.exe /c "R.exe -q -e ""install.packages($DEPS, repos='$CRAN', type='both')"" 2>&1"
           $BINARY_DEPS = "c('XML','igraph')"
           cmd.exe /c "R.exe -q -e ""install.packages($BINARY_DEPS, repos='$CRAN', type='win.binary')"" 2>&1"
+          $DEPS = "c('data.table','magrittr','stringi','ggplot2','DiagrammeR','Ckmeans.1d.dp','vcd','testthat','lintr','knitr','rmarkdown')"
+          cmd.exe /c "R.exe -q -e ""install.packages($DEPS, repos='$CRAN', type='both')"" 2>&1"
         }
 
 build_script:
@@ -96,10 +102,11 @@ build_script:
 
 test_script:
     - cd %APPVEYOR_BUILD_FOLDER%
-    - if /i "%DO_PYTHON%" == "on" python -m nose tests/python
+    - if /i "%DO_PYTHON%" == "on" python -m pytest tests/python
     # mingw R package: run the R check (which includes unit tests), and also keep the built binary package
     - if /i "%target%" == "rmingw" (
         set _R_CHECK_CRAN_INCOMING_=FALSE&&
+        set _R_CHECK_FORCE_SUGGESTS_=FALSE&&
         R.exe CMD check xgboost*.tar.gz --no-manual --no-build-vignettes --as-cran --install-args=--build
       )
     # MSVC R package: run only the unit tests

build.sh

@@ -1,51 +0,0 @@
-#!/bin/bash
-# This is a simple script to make xgboost in MAC and Linux
-# Basically, it first try to make with OpenMP, if fails, disable OpenMP and make it again.
-# This will automatically make xgboost for MAC users who don't have OpenMP support.
-# In most cases, type make will give what you want.
-
-# See additional instruction in doc/build.md
-set -e
-
-if make; then
-    echo "Successfully build multi-thread xgboost"
-else
-
-    not_ready=0
-
-    if [[ ! -e ./rabit/Makefile ]]; then
-        echo ""
-        echo "Please init the rabit submodule:"
-        echo "git submodule update --init --recursive -- rabit"
-        not_ready=1
-    fi
-    
-    if [[ ! -e ./dmlc-core/Makefile ]]; then
-        echo ""
-        echo "Please init the dmlc-core submodule:"
-        echo "git submodule update --init --recursive -- dmlc-core"
-        not_ready=1
-    fi
-
-    if [[ "${not_ready}" == "1" ]]; then
-        echo ""
-        echo "Please fix the errors above and retry the build, or reclone the repository with:"
-        echo "git clone --recursive https://github.com/dmlc/xgboost.git"
-        echo ""
-        exit 1
-    fi
-
-
-    echo "-----------------------------"
-    echo "Building multi-thread xgboost failed"
-    echo "Start to build single-thread xgboost"
-    make clean_all
-    make config=make/minimum.mk
-    if [ $? -eq 0 ] ;then
-      echo "Successfully build single-thread xgboost"
-      echo "If you want multi-threaded version"
-      echo "See additional instructions in doc/build.md"
-    else
-      echo "Failed to build single-thread xgboost"
-    fi
-fi

cmake/Doc.cmake

@@ -0,0 +1,16 @@
+function (run_doxygen)
+  find_package(Doxygen REQUIRED)
+
+  if (NOT DOXYGEN_DOT_FOUND)
+    message(FATAL_ERROR "Command `dot` not found.  Please install graphviz.")
+  endif (NOT DOXYGEN_DOT_FOUND)
+
+  configure_file(
+    ${PROJECT_SOURCE_DIR}/doc/Doxyfile.in
+    ${CMAKE_CURRENT_BINARY_DIR}/Doxyfile @ONLY)
+  add_custom_target( doc_doxygen ALL
+    COMMAND ${DOXYGEN_EXECUTABLE} ${CMAKE_CURRENT_BINARY_DIR}/Doxyfile
+    WORKING_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}
+    COMMENT "Generate C APIs documentation."
+    VERBATIM)
+endfunction (run_doxygen)

cmake/Utils.cmake

@@ -1,4 +1,3 @@
-
 # Automatically set source group based on folder
 function(auto_source_group SOURCES)
 
@@ -18,6 +17,10 @@ endfunction(auto_source_group)
 function(msvc_use_static_runtime)
   if(MSVC)
       set(variables
+          CMAKE_C_FLAGS_DEBUG
+          CMAKE_C_FLAGS_MINSIZEREL
+          CMAKE_C_FLAGS_RELEASE
+          CMAKE_C_FLAGS_RELWITHDEBINFO
           CMAKE_CXX_FLAGS_DEBUG
           CMAKE_CXX_FLAGS_MINSIZEREL
           CMAKE_CXX_FLAGS_RELEASE
@@ -29,25 +32,42 @@ function(msvc_use_static_runtime)
               set(${variable} "${${variable}}"  PARENT_SCOPE)
           endif()
       endforeach()
+      set(variables
+          CMAKE_CUDA_FLAGS
+          CMAKE_CUDA_FLAGS_DEBUG
+          CMAKE_CUDA_FLAGS_MINSIZEREL
+          CMAKE_CUDA_FLAGS_RELEASE
+          CMAKE_CUDA_FLAGS_RELWITHDEBINFO
+      )
+      foreach(variable ${variables})
+          if(${variable} MATCHES "-MD")
+              string(REGEX REPLACE "-MD" "-MT" ${variable} "${${variable}}")
+              set(${variable} "${${variable}}"  PARENT_SCOPE)
+          endif()
+          if(${variable} MATCHES "/MD")
+              string(REGEX REPLACE "/MD" "/MT" ${variable} "${${variable}}")
+              set(${variable} "${${variable}}"  PARENT_SCOPE)
+          endif()
+      endforeach()
   endif()
 endfunction(msvc_use_static_runtime)
 
 # Set output directory of target, ignoring debug or release
 function(set_output_directory target dir)
-	set_target_properties(${target} PROPERTIES 
+	set_target_properties(${target} PROPERTIES
 		RUNTIME_OUTPUT_DIRECTORY ${dir}
 		RUNTIME_OUTPUT_DIRECTORY_DEBUG ${dir}
 		RUNTIME_OUTPUT_DIRECTORY_RELEASE ${dir}
-		LIBRARY_OUTPUT_DIRECTORY ${dir} 
-		LIBRARY_OUTPUT_DIRECTORY_DEBUG ${dir} 
-		LIBRARY_OUTPUT_DIRECTORY_RELEASE ${dir} 
+		LIBRARY_OUTPUT_DIRECTORY ${dir}
+		LIBRARY_OUTPUT_DIRECTORY_DEBUG ${dir}
+		LIBRARY_OUTPUT_DIRECTORY_RELEASE ${dir}
 	)
 endfunction(set_output_directory)
 
 # Set a default build type to release if none was specified
 function(set_default_configuration_release)
     if(CMAKE_CONFIGURATION_TYPES STREQUAL "Debug;Release;MinSizeRel;RelWithDebInfo") # multiconfig generator?
-        set(CMAKE_CONFIGURATION_TYPES Release CACHE STRING "" FORCE) 
+        set(CMAKE_CONFIGURATION_TYPES Release CACHE STRING "" FORCE)
 	elseif(NOT CMAKE_BUILD_TYPE AND NOT CMAKE_CONFIGURATION_TYPES)
 	  message(STATUS "Setting build type to 'Release' as none was specified.")
 	  set(CMAKE_BUILD_TYPE Release CACHE STRING "Choose the type of build." FORCE )
@@ -57,9 +77,14 @@ endfunction(set_default_configuration_release)
 # Generate nvcc compiler flags given a list of architectures
 # Also generates PTX for the most recent architecture for forwards compatibility
 function(format_gencode_flags flags out)
+  if(CMAKE_CUDA_COMPILER_VERSION MATCHES "^([0-9]+\\.[0-9]+)")
+    set(CUDA_VERSION "${CMAKE_MATCH_1}")
+  endif()
   # Set up architecture flags
-  if(NOT flags) 
-    if((CUDA_VERSION_MAJOR EQUAL 9) OR (CUDA_VERSION_MAJOR GREATER 9))
+  if(NOT flags)
+    if(CUDA_VERSION VERSION_GREATER_EQUAL "10.0")
+      set(flags "35;50;52;60;61;70;75")
+    elseif(CUDA_VERSION VERSION_GREATER_EQUAL "9.0")
       set(flags "35;50;52;60;61;70")
     else()
       set(flags "35;50;52;60;61")
@@ -67,12 +92,12 @@ function(format_gencode_flags flags out)
   endif()
   # Generate SASS
   foreach(ver ${flags})
-    set(${out} "${${out}}-gencode arch=compute_${ver},code=sm_${ver};")
+    set(${out} "${${out}}--generate-code=arch=compute_${ver},code=sm_${ver};")
   endforeach()
   # Generate PTX for last architecture
   list(GET flags -1 ver)
-  set(${out} "${${out}}-gencode arch=compute_${ver},code=compute_${ver};")
-  
+  set(${out} "${${out}}--generate-code=arch=compute_${ver},code=compute_${ver};")
+
   set(${out} "${${out}}" PARENT_SCOPE)
 endfunction(format_gencode_flags flags)
 
@@ -80,6 +105,10 @@ endfunction(format_gencode_flags flags)
 # if necessary, installs the main R package dependencies;
 # runs R CMD INSTALL.
 function(setup_rpackage_install_target rlib_target build_dir)
+  # backup cmake_install.cmake
+  install(CODE "file(COPY \"${build_dir}/R-package/cmake_install.cmake\"
+DESTINATION \"${build_dir}/bak\")")
+
   install(CODE "file(REMOVE_RECURSE \"${build_dir}/R-package\")")
   install(
     DIRECTORY "${PROJECT_SOURCE_DIR}/R-package"
@@ -98,4 +127,8 @@ function(setup_rpackage_install_target rlib_target build_dir)
   install(CODE "execute_process(COMMAND \"${LIBR_EXECUTABLE}\" \"-q\" \"-e\" \"${XGB_DEPS_SCRIPT}\")")
   install(CODE "execute_process(COMMAND \"${LIBR_EXECUTABLE}\" CMD INSTALL\
     \"--no-multiarch\" \"--build\" \"${build_dir}/R-package\")")
+
+  # restore cmake_install.cmake
+  install(CODE "file(RENAME \"${build_dir}/bak/cmake_install.cmake\"
+ \"${build_dir}/R-package/cmake_install.cmake\")")
 endfunction(setup_rpackage_install_target)

cmake/modules/FindASan.cmake

@@ -1,8 +1,8 @@
 set(ASan_LIB_NAME ASan)
 
 find_library(ASan_LIBRARY
-  NAMES libasan.so libasan.so.4
-  PATHS /usr/lib64 /usr/lib /usr/local/lib64 /usr/local/lib)
+  NAMES libasan.so libasan.so.4 libasan.so.3 libasan.so.2 libasan.so.1 libasan.so.0
+  PATHS ${SANITIZER_PATH} /usr/lib64 /usr/lib /usr/local/lib64 /usr/local/lib ${CMAKE_PREFIX_PATH}/lib)
 
 include(FindPackageHandleStandardArgs)
 find_package_handle_standard_args(ASan DEFAULT_MSG

cmake/modules/FindLSan.cmake

@@ -2,7 +2,7 @@ set(LSan_LIB_NAME lsan)
 
 find_library(LSan_LIBRARY
   NAMES liblsan.so liblsan.so.0 liblsan.so.0.0.0
-  PATHS /usr/lib64 /usr/lib /usr/local/lib64 /usr/local/lib)
+  PATHS ${SANITIZER_PATH} /usr/lib64 /usr/lib /usr/local/lib64 /usr/local/lib ${CMAKE_PREFIX_PATH}/lib)
 
 include(FindPackageHandleStandardArgs)
 find_package_handle_standard_args(LSan DEFAULT_MSG

cmake/modules/FindNccl.cmake

@@ -32,20 +32,28 @@
 #
 # This module assumes that the user has already called find_package(CUDA)
 
+if (NCCL_LIBRARY)
+  # Don't cache NCCL_LIBRARY to enable switching between static and shared.
+  unset(NCCL_LIBRARY CACHE)
+endif()
 
-set(NCCL_LIB_NAME nccl_static)
+if (BUILD_WITH_SHARED_NCCL)
+  # libnccl.so
+  set(NCCL_LIB_NAME nccl)
+else ()
+  # libnccl_static.a
+  set(NCCL_LIB_NAME nccl_static)
+endif (BUILD_WITH_SHARED_NCCL)
 
 find_path(NCCL_INCLUDE_DIR
   NAMES nccl.h
-  PATHS $ENV{NCCL_ROOT}/include ${NCCL_ROOT}/include ${CUDA_INCLUDE_DIRS} /usr/include)
+  PATHS $ENV{NCCL_ROOT}/include ${NCCL_ROOT}/include)
 
 find_library(NCCL_LIBRARY
   NAMES ${NCCL_LIB_NAME}
-  PATHS $ENV{NCCL_ROOT}/lib ${NCCL_ROOT}/lib ${CUDA_INCLUDE_DIRS}/../lib /usr/lib)
+  PATHS $ENV{NCCL_ROOT}/lib/ ${NCCL_ROOT}/lib)
 
-if (NCCL_INCLUDE_DIR AND NCCL_LIBRARY)
-  get_filename_component(NCCL_LIBRARY ${NCCL_LIBRARY} PATH)
-endif ()
+message(STATUS "Using nccl library: ${NCCL_LIBRARY}")
 
 include(FindPackageHandleStandardArgs)
 find_package_handle_standard_args(Nccl DEFAULT_MSG
@@ -54,5 +62,4 @@ find_package_handle_standard_args(Nccl DEFAULT_MSG
 mark_as_advanced(
   NCCL_INCLUDE_DIR
   NCCL_LIBRARY
-  NCCL_LIB_NAME
 )

cmake/modules/FindTSan.cmake

@@ -2,7 +2,7 @@ set(TSan_LIB_NAME tsan)
 
 find_library(TSan_LIBRARY
   NAMES libtsan.so libtsan.so.0 libtsan.so.0.0.0
-  PATHS /usr/lib64 /usr/lib /usr/local/lib64 /usr/local/lib)
+  PATHS ${SANITIZER_PATH} /usr/lib64 /usr/lib /usr/local/lib64 /usr/local/lib ${CMAKE_PREFIX_PATH}/lib)
 
 include(FindPackageHandleStandardArgs)
 find_package_handle_standard_args(TSan DEFAULT_MSG

cmake/xgboost-config.cmake.in

@@ -0,0 +1,5 @@
+@PACKAGE_INIT@
+
+if(NOT TARGET xgboost::xgboost)
+  include(${CMAKE_CURRENT_LIST_DIR}/XGBoostTargets.cmake)
+endif()

demo/README.md

@@ -119,6 +119,7 @@ If you have particular usecase of xgboost that you would like to highlight.
 Send a PR to add a one sentence description:)
 
 - XGBoost is used in [Kaggle Script](https://www.kaggle.com/scripts) to solve data science challenges.
+- Distribute XGBoost as Rest API server from Jupyter notebook with [BentoML](https://github.com/bentoml/bentoml). [Link to notebook](https://github.com/bentoml/BentoML/blob/master/examples/xgboost-predict-titanic-survival/XGBoost-titanic-survival-prediction.ipynb)
 - [Seldon predictive service powered by XGBoost](http://docs.seldon.io/iris-demo.html)
 - XGBoost Distributed is used in [ODPS Cloud Service by Alibaba](https://yq.aliyun.com/articles/6355) (in Chinese)
 - XGBoost is incoporated as part of [Graphlab Create](https://dato.com/products/create/) for scalable machine learning.
@@ -135,6 +136,7 @@ Send a PR to add a one sentence description:)
 
 ## Awards
 - [John Chambers Award](http://stat-computing.org/awards/jmc/winners.html) - 2016 Winner: XGBoost R Package, by Tong He (Simon Fraser University) and Tianqi Chen (University of Washington)
+- [InfoWorld’s 2019 Technology of the Year Award](https://www.infoworld.com/article/3336072/application-development/infoworlds-2019-technology-of-the-year-award-winners.html)
 
 ## Windows Binaries
 Unofficial windows binaries and instructions on how to use them are hosted on [Guido Tapia's blog](http://www.picnet.com.au/blogs/guido/post/2016/09/22/xgboost-windows-x64-binaries-for-download/)

demo/binary_classification/README.md

@@ -62,7 +62,7 @@ test:data = "agaricus.txt.test"
 We use the tree booster and logistic regression objective in our setting. This indicates that we accomplish our task using classic gradient boosting regression tree(GBRT), which is a promising method for binary classification.
 
 The parameters shown in the example gives the most common ones that are needed to use xgboost.
-If you are interested in more parameter settings, the complete parameter settings and detailed descriptions are [here](../../doc/parameter.md). Besides putting the parameters in the configuration file, we can set them by passing them as arguments as below:
+If you are interested in more parameter settings, the complete parameter settings and detailed descriptions are [here](../../doc/parameter.rst). Besides putting the parameters in the configuration file, we can set them by passing them as arguments as below:
 
 ```
 ../../xgboost mushroom.conf max_depth=6
@@ -80,12 +80,6 @@ booster = gblinear
 # L2 regularization term on weights, default 0
 lambda = 0.01
 # L1 regularization term on weights, default 0
-If ```agaricus.txt.test.buffer``` exists, and automatically loads from binary buffer if possible, this can speedup training process when you do training many times. You can disable it by setting ```use_buffer=0```.
-  - Buffer file can also be used as standalone input, i.e if buffer file exists, but original agaricus.txt.test was removed, xgboost will still run
-* Deviation from LibSVM input format: xgboost is compatible with LibSVM format, with the following minor differences:
-  - xgboost allows feature index starts from 0
-  - for binary classification, the label is 1 for positive, 0 for negative, instead of +1,-1
-  - the feature indices in each line *do not* need to be sorted
 alpha = 0.01
 # L2 regularization term on bias, default 0
 lambda_bias = 0.01
@@ -102,7 +96,7 @@ After training, we can use the output model to get the prediction of the test da
 For binary classification, the output predictions are probability confidence scores in [0,1], corresponds to the probability of the label to be positive.
 
 #### Dump Model
-This is a preliminary feature, so far only tree model support text dump. XGBoost can display the tree models in text files and we can scan the model in an easy way:
+This is a preliminary feature, so only tree models support text dump. XGBoost can display the tree models in text or JSON files, and we can scan the model in an easy way:
 ```
 ../../xgboost mushroom.conf task=dump model_in=0002.model name_dump=dump.raw.txt
 ../../xgboost mushroom.conf task=dump model_in=0002.model fmap=featmap.txt name_dump=dump.nice.txt

demo/binary_classification/mapfeat.py

@@ -18,7 +18,7 @@ def loadfmap( fname ):
             if it.strip() == '':
                 continue
             k , v = it.split('=')
-            fmap[ idx ][ v ] = len(nmap) + 1
+            fmap[ idx ][ v ] = len(nmap)
             nmap[ len(nmap) ] = ftype+'='+k
     return fmap, nmap
 

demo/c-api/Makefile

@@ -0,0 +1,19 @@
+SRC=c-api-demo.c
+TGT=c-api-demo
+
+cc=cc
+CFLAGS ?=-O3
+XGBOOST_ROOT ?=../..
+INCLUDE_DIR=-I$(XGBOOST_ROOT)/include -I$(XGBOOST_ROOT)/dmlc-core/include -I$(XGBOOST_ROOT)/rabit/include
+LIB_DIR=-L$(XGBOOST_ROOT)/lib
+
+build: $(TGT)
+
+$(TGT): $(SRC) Makefile
+	$(cc) $(CFLAGS) $(INCLUDE_DIR) $(LIB_DIR) -o $(TGT) $(SRC) -lxgboost
+
+run: $(TGT)
+	LD_LIBRARY_PATH=$(XGBOOST_ROOT)/lib ./$(TGT)
+
+clean:
+	rm -f $(TGT)

demo/c-api/README.md

@@ -0,0 +1,30 @@
+C-APIs
+===
+
+**XGBoost** implements a C API originally designed for various language
+bindings.  For detailed reference, please check xgboost/c_api.h.  Here is a
+demonstration of using the API.
+
+# CMake
+If you use **CMake** for your project, you can either install **XGBoost**
+somewhere in your system and tell CMake to find it by calling
+`find_package(xgboost)`, or put **XGBoost** inside your project's source tree
+and call **CMake** command: `add_subdirectory(xgboost)`.  To use
+`find_package()`, put the following in your **CMakeLists.txt**:
+
+``` CMake
+find_package(xgboost REQUIRED)
+add_executable(api-demo c-api-demo.c)
+target_link_libraries(api-demo xgboost::xgboost)
+```
+
+If you want to put XGBoost inside your project (like git submodule), use this
+instead:
+``` CMake
+add_subdirectory(xgboost)
+add_executable(api-demo c-api-demo.c)
+target_link_libraries(api-demo xgboost)
+```
+
+# make
+You can start by modifying the makefile in this directory to fit your need.

demo/c-api/c-api-demo.c

@@ -0,0 +1,89 @@
+/*!
+ * Copyright 2019 XGBoost contributors
+ *
+ * \file c-api-demo.c
+ * \brief A simple example of using xgboost C API.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <xgboost/c_api.h>
+
+#define safe_xgboost(call) {                                            \
+int err = (call);                                                       \
+if (err != 0) {                                                         \
+  fprintf(stderr, "%s:%d: error in %s: %s\n", __FILE__, __LINE__, #call, XGBGetLastError()); \
+  exit(1);                                                              \
+}                                                                       \
+}
+
+int main(int argc, char** argv) {
+  int silent = 0;
+  int use_gpu = 0;  // set to 1 to use the GPU for training
+  
+  // load the data
+  DMatrixHandle dtrain, dtest;
+  safe_xgboost(XGDMatrixCreateFromFile("../data/agaricus.txt.train", silent, &dtrain));
+  safe_xgboost(XGDMatrixCreateFromFile("../data/agaricus.txt.test", silent, &dtest));
+  
+  // create the booster
+  BoosterHandle booster;
+  DMatrixHandle eval_dmats[2] = {dtrain, dtest};
+  safe_xgboost(XGBoosterCreate(eval_dmats, 2, &booster));
+
+  // configure the training
+  // available parameters are described here:
+  //   https://xgboost.readthedocs.io/en/latest/parameter.html
+  safe_xgboost(XGBoosterSetParam(booster, "tree_method", use_gpu ? "gpu_hist" : "hist"));
+  if (use_gpu) {
+    // set the number of GPUs and the first GPU to use;
+    // this is not necessary, but provided here as an illustration
+    safe_xgboost(XGBoosterSetParam(booster, "n_gpus", "1"));
+    safe_xgboost(XGBoosterSetParam(booster, "gpu_id", "0"));
+  } else {
+    // avoid evaluating objective and metric on a GPU
+    safe_xgboost(XGBoosterSetParam(booster, "n_gpus", "0"));
+  }
+
+  safe_xgboost(XGBoosterSetParam(booster, "objective", "binary:logistic"));
+  safe_xgboost(XGBoosterSetParam(booster, "min_child_weight", "1"));
+  safe_xgboost(XGBoosterSetParam(booster, "gamma", "0.1"));
+  safe_xgboost(XGBoosterSetParam(booster, "max_depth", "3"));
+  safe_xgboost(XGBoosterSetParam(booster, "verbosity", silent ? "0" : "1"));
+  
+  // train and evaluate for 10 iterations
+  int n_trees = 10;
+  const char* eval_names[2] = {"train", "test"};
+  const char* eval_result = NULL;
+  for (int i = 0; i < n_trees; ++i) {
+    safe_xgboost(XGBoosterUpdateOneIter(booster, i, dtrain));
+    safe_xgboost(XGBoosterEvalOneIter(booster, i, eval_dmats, eval_names, 2, &eval_result));
+    printf("%s\n", eval_result);
+  }
+
+  // predict
+  bst_ulong out_len = 0;
+  const float* out_result = NULL;
+  int n_print = 10;
+
+  safe_xgboost(XGBoosterPredict(booster, dtest, 0, 0, &out_len, &out_result));
+  printf("y_pred: ");
+  for (int i = 0; i < n_print; ++i) {
+    printf("%1.4f ", out_result[i]);
+  }
+  printf("\n");
+
+  // print true labels
+  safe_xgboost(XGDMatrixGetFloatInfo(dtest, "label", &out_len, &out_result));
+  printf("y_test: ");
+  for (int i = 0; i < n_print; ++i) {
+    printf("%1.4f ", out_result[i]);
+  }
+  printf("\n");
+
+  // free everything
+  safe_xgboost(XGBoosterFree(booster));
+  safe_xgboost(XGDMatrixFree(dtrain));
+  safe_xgboost(XGDMatrixFree(dtest));
+  return 0;
+}