Commit Graph
163 Commits
Author SHA1 Message Date
AbdealiJK b94fcab4dc Add dump_format=json option (#1726)
* Add format to the params accepted by DumpModel

Currently, only the test format is supported when trying to dump
a model. The plan is to add more such formats like JSON which are
easy to read and/or parse by machines. And to make the interface
for this even more generic to allow other formats to be added.

Hence, we make some modifications to make these function generic
and accept a new parameter "format" which signifies the format of
the dump to be created.

* Fix typos and errors in docs

* plugin: Mention all the register macros available

Document the register macros currently available to the plugin
writers so they know what exactly can be extended using hooks.

* sparce_page_source: Use same arg name in .h and .cc

* gbm: Add JSON dump

The dump_format argument can be used to specify what type
of dump file should be created. Add functionality to dump
gblinear and gbtree into a JSON file.

The JSON file has an array, each item is a JSON object for the tree.
For gblinear:
 - The item is the bias and weights vectors
For gbtree:
 - The item is the root node. The root node has a attribute "children"
   which holds the children nodes. This happens recursively.

* core.py: Add arg dump_format for get_dump()
2016-11-04 09:55:25 -07:00
Tianqi Chen c93c9b7ed6 [TREE] Experimental version of monotone constraint (#1516)
* [TREE] Experimental version of monotone constraint

* Allow default detection of montone option

* loose the condition of strict check

* Update gbtree.cc
2016-09-07 21:28:43 -07:00
Tianqi Chen ecec5f7959 [CORE] Refactor cache mechanism (#1540) 2016-09-02 20:39:07 -07:00
Vadim Khotilovich 75f401481f no exception throwing within omp parallel; set nthread in Learner (#1421) 2016-07-29 10:08:03 -07:00
anpark 0e61c514a7 fix duplicate loop over output_group when predict (#1342)
* fix sparse page source meta info empty when load from dmatrix

* fix duplicate loop over output_group when predict
2016-07-13 10:03:10 -07:00
Yoshinori Nakano 7cfeb5f012 fix Dart::NormalizeTrees (#1265) 2016-06-09 15:28:24 -07:00
Yoshinori Nakano 949d1e3027 add Dart booster (#1220) 2016-06-08 14:04:01 -07:00
tqchen 96b17971ac Fix continue training in CLI 2016-03-10 12:43:25 -08:00
tqchen e80d3db64b [DIST] Enable multiple thread and tracker, make rabit and xgboost more thread-safe by using thread local variables. 2016-03-03 20:36:14 -08:00
tqchen 88447ca32e [MEM] Add rowset struct to save memory with billion level rows 2016-02-10 11:17:17 -08:00
tqchen d75e3ed05d [LIBXGBOOST] pass demo running. 2016-01-16 10:24:01 -08:00
tqchen 0d95e863c9 [LEARNER] refactor learner 2016-01-16 10:24:01 -08:00
tqchen 4b4b36d047 [GBM] remove need to explicit InitModel, rename save/load 2016-01-16 10:24:01 -08:00
tqchen 82ceb4de0a [LEARNER] Init learner interface 2016-01-16 10:24:01 -08:00
tqchen 9042b9e2c7 [GBM] Finish migrate all gbms 2016-01-16 10:24:01 -08:00
tqchen 4f26d98150 [Update] remove rabit subtree, use submodule, move code 2016-01-16 10:24:01 -08:00
tqchen d530e0c14f [REFACTOR] cleanup structure 2016-01-16 10:24:00 -08:00
Vadim Khotilovich c70022e6c4 spelling, wording, and doc fixes in c++ code
I was reading through the code and fixing some things in the comments.
Only a few trivial actual code changes were made to make things more
readable.
2015-12-12 21:40:12 -06:00
yoori 981f06b9d1 style fix 2015-10-20 00:58:11 +04:00
yoori 49c1cb6990 GBTree::Predict performance fix: removed excess thread_temp initialization 2015-10-20 00:52:37 +04:00
yoori c0853967d5 GBTree::Predict performance fix: removed excess thread_temp initialization 2015-10-20 00:06:00 +04:00
tqchen 0162bb7034 lint half way 2015-07-03 18:31:52 -07:00
tqchen 7f7947f31c add with pbuffer info to model, allow xgb model to be saved in a more memory compact way 2015-05-06 15:43:15 -07:00
Tianqi Chen afdebe8d8f fix platform dependent thing 2015-04-25 20:40:43 -07:00
tqchen fba9e5c714 quick fix 2015-04-05 12:01:19 -07:00
tqchen 12528c535a fix 2015-03-11 11:22:51 -07:00
tqchen a16cbedfab try fix memleak when test data have more features than training 2015-02-10 21:49:29 -08:00
tqchen 1211ea40c9 add single instance prediction 2015-01-19 08:07:22 -08:00
tqchen b762231b02 change makefile to lazy checkpt, fix col splt code 2015-01-15 21:32:31 -08:00
tqchen c8f422b3b9 add dump to linear model 2014-12-24 02:56:32 -08:00
tqchen 31eedfea59 pas mock, need to fix rabit lib for not initialization 2014-12-21 00:14:00 -08:00
tqchen deb21351b9 add rabit checkpoint to xgb 2014-12-20 01:05:40 -08:00
tqchen db2adb6885 start check windows compatiblity 2014-11-23 20:59:10 -08:00
Tianqi Chen 2e444f8338 remove warning from MSVC need another round of check 2014-11-23 20:52:13 -08:00
tqchen 168bb0d0c9 add predict leaf indices 2014-11-21 09:32:09 -08:00
tqchen 970dd58dc2 checkin continue training 2014-11-19 20:06:08 -08:00
tqchen 0cf2dd39ea new change for mpi 2014-10-16 15:12:10 -07:00
tqchen d4ab359be1 fix 2014-09-07 20:01:03 -07:00
tqchen 19a1ee24a5 try predpath 2014-09-07 18:40:15 -07:00
tqchen 10648a1ca7 remove using std from cpp 2014-09-02 22:43:19 -07:00
tqchen 9100ffc12a chg version 2014-09-01 22:32:03 -07:00
tqchen 4592e500cb add ntree limit 2014-09-01 15:10:19 -07:00
tqchen@graphlab.com 075dc9a998 pass build 2014-08-27 19:19:04 -07:00
tqchen@graphlab.com 605269133e complete refactor data.h, now replies on iterator to access column 2014-08-27 17:00:21 -07:00
tqchen@graphlab.com a59f8945dc rename SparseBatch to RowBatch 2014-08-27 10:56:55 -07:00
tqchen@graphlab.com d5a5e0a42a rename findex->index 2014-08-27 10:52:27 -07:00
tqchen 414e7f27ff Merge branch 'master' into unity
Conflicts:
	src/learner/evaluation-inl.hpp
	wrapper/xgboost_R.cpp
	wrapper/xgboost_wrapper.cpp
	wrapper/xgboost_wrapper.h
2014-08-26 20:32:07 -07:00
tqchen 7739f57c8b change omp loop var to bst_omp_uint, add XGB_DLL to wrapper 2014-08-26 19:37:04 -07:00
tqchen@graphlab.com d61b0b757f chg 2014-08-25 11:35:38 -07:00
tqchen@graphlab.com c78a2164c2 fix line from auto spacing by msvc 2014-08-25 11:34:49 -07:00