Support categorical split in tree model dump. (#7036)

include/xgboost/feature_map.h

@@ -1,5 +1,5 @@
 /*!
- * Copyright 2014 by Contributors
+ * Copyright 2014-2021 by Contributors
  * \file feature_map.h
  * \brief Feature map data structure to help visualization and model dump.
  * \author Tianqi Chen
@@ -26,7 +26,8 @@ class FeatureMap {
     kIndicator = 0,
     kQuantitive = 1,
     kInteger = 2,
-    kFloat = 3
+    kFloat = 3,
+    kCategorical = 4
   };
   /*!
    * \brief load feature map from input stream
@@ -82,6 +83,7 @@ class FeatureMap {
     if (!strcmp("q", tname)) return kQuantitive;
     if (!strcmp("int", tname)) return kInteger;
     if (!strcmp("float", tname)) return kFloat;
+    if (!strcmp("categorical", tname)) return kCategorical;
     LOG(FATAL) << "unknown feature type, use i for indicator and q for quantity";
     return kIndicator;
   }

python-package/xgboost/plotting.py

@@ -3,6 +3,7 @@
 # coding: utf-8
 """Plotting Library."""
 from io import BytesIO
+import json
 import numpy as np
 from .core import Booster
 from .sklearn import XGBModel
@@ -203,7 +204,7 @@ def to_graphviz(booster, fmap='', num_trees=0, rankdir=None,
 
     if kwargs:
         parameters += ':'
-        parameters += str(kwargs)
+        parameters += json.dumps(kwargs)
     tree = booster.get_dump(
         fmap=fmap,
         dump_format=parameters)[num_trees]

src/predictor/cpu_predictor.cc

@@ -52,11 +52,6 @@ bst_float PredValue(const SparsePage::Inst &inst,
     if (tree_info[i] == bst_group) {
       auto const &tree = *trees[i];
       bool has_categorical = tree.HasCategoricalSplit();
-
-      auto categories = common::Span<uint32_t const>{tree.GetSplitCategories()};
-      auto split_types = tree.GetSplitTypes();
-      auto categories_ptr =
-          common::Span<RegTree::Segment const>{tree.GetSplitCategoriesPtr()};
       auto cats = tree.GetCategoriesMatrix();
       bst_node_t nidx = -1;
       if (has_categorical) {

src/tree/tree_model.cc

@@ -1,5 +1,5 @@
 /*!
- * Copyright 2015-2020 by Contributors
+ * Copyright 2015-2021 by Contributors
  * \file tree_model.cc
  * \brief model structure for tree
  */
@@ -74,6 +74,7 @@ class TreeGenerator {
                                 int32_t /*nid*/, uint32_t /*depth*/) const {
     return "";
   }
+  virtual std::string Categorical(RegTree const&, int32_t, uint32_t) const = 0;
   virtual std::string Integer(RegTree const& /*tree*/,
                                 int32_t /*nid*/, uint32_t /*depth*/) const {
     return "";
@@ -92,27 +93,52 @@ class TreeGenerator {
   virtual std::string SplitNode(RegTree const& tree, int32_t nid, uint32_t depth) {
     auto const split_index = tree[nid].SplitIndex();
     std::string result;
+    auto is_categorical = tree.GetSplitTypes()[nid] == FeatureType::kCategorical;
     if (split_index < fmap_.Size()) {
+      auto check_categorical = [&]() {
+        CHECK(is_categorical)
+            << fmap_.Name(split_index)
+            << " in feature map is numerical but tree node is categorical.";
+      };
+      auto check_numerical = [&]() {
+        auto is_numerical = !is_categorical;
+        CHECK(is_numerical)
+            << fmap_.Name(split_index)
+            << " in feature map is categorical but tree node is numerical.";
+      };
+
       switch (fmap_.TypeOf(split_index)) {
+      case FeatureMap::kCategorical: {
+        check_categorical();
+        result = this->Categorical(tree, nid, depth);
+        break;
+      }
       case FeatureMap::kIndicator: {
+        check_numerical();
         result = this->Indicator(tree, nid, depth);
         break;
       }
       case FeatureMap::kInteger: {
+        check_numerical();
         result = this->Integer(tree, nid, depth);
         break;
       }
       case FeatureMap::kFloat:
       case FeatureMap::kQuantitive: {
+        check_numerical();
         result = this->Quantitive(tree, nid, depth);
         break;
       }
       default:
         LOG(FATAL) << "Unknown feature map type.";
       }
+    } else {
+      if (is_categorical) {
+        result = this->Categorical(tree, nid, depth);
       } else {
         result = this->PlainNode(tree, nid, depth);
       }
+    }
     return result;
   }
 
@@ -179,6 +205,32 @@ TreeGenerator* TreeGenerator::Create(std::string const& attrs, FeatureMap const&
   __make_ ## TreeGenReg ## _ ## UniqueId ## __ =                        \
                   ::dmlc::Registry< ::xgboost::TreeGenReg>::Get()->__REGISTER__(Name)
 
+std::vector<bst_cat_t> GetSplitCategories(RegTree const &tree, int32_t nidx) {
+  auto const &csr = tree.GetCategoriesMatrix();
+  auto seg = csr.node_ptr[nidx];
+  auto split = common::KCatBitField{csr.categories.subspan(seg.beg, seg.size)};
+
+  std::vector<bst_cat_t> cats;
+  for (size_t i = 0; i < split.Size(); ++i) {
+    if (split.Check(i)) {
+      cats.push_back(static_cast<bst_cat_t>(i));
+    }
+  }
+  return cats;
+}
+
+std::string PrintCatsAsSet(std::vector<bst_cat_t> const &cats) {
+  std::stringstream ss;
+  ss << "{";
+  for (size_t i = 0; i < cats.size(); ++i) {
+    ss << cats[i];
+    if (i != cats.size() - 1) {
+      ss << ",";
+    }
+  }
+  ss << "}";
+  return ss.str();
+}
 
 class TextGenerator : public TreeGenerator {
   using SuperT = TreeGenerator;
@@ -258,6 +310,17 @@ class TextGenerator : public TreeGenerator {
     return SplitNodeImpl(tree, nid, kNodeTemplate, SuperT::ToStr(cond), depth);
   }
 
+  std::string Categorical(RegTree const &tree, int32_t nid,
+                       uint32_t depth) const override {
+    auto cats = GetSplitCategories(tree, nid);
+    std::string cats_str = PrintCatsAsSet(cats);
+    static std::string const kNodeTemplate =
+        "{tabs}{nid}:[{fname}:{cond}] yes={right},no={left},missing={missing}";
+    std::string const result =
+        SplitNodeImpl(tree, nid, kNodeTemplate, cats_str, depth);
+    return result;
+  }
+
   std::string NodeStat(RegTree const& tree, int32_t nid) const override {
     static std::string const kStatTemplate = ",gain={loss_chg},cover={sum_hess}";
     std::string const result = SuperT::Match(
@@ -343,6 +406,24 @@ class JsonGenerator : public TreeGenerator {
     return result;
   }
 
+  std::string Categorical(RegTree const& tree, int32_t nid, uint32_t depth) const override {
+    auto cats = GetSplitCategories(tree, nid);
+    static std::string const kCategoryTemplate =
+        R"I( "nodeid": {nid}, "depth": {depth}, "split": "{fname}", )I"
+        R"I("split_condition": {cond}, "yes": {right}, "no": {left}, )I"
+        R"I("missing": {missing})I";
+    std::string cats_ptr = "[";
+    for (size_t i = 0; i < cats.size(); ++i) {
+      cats_ptr += std::to_string(cats[i]);
+      if (i != cats.size() - 1) {
+        cats_ptr += ", ";
+      }
+    }
+    cats_ptr += "]";
+    auto results = SplitNodeImpl(tree, nid, kCategoryTemplate, cats_ptr, depth);
+    return results;
+  }
+
   std::string SplitNodeImpl(RegTree const &tree, int32_t nid,
                             std::string const &template_str, std::string cond,
                             uint32_t depth) const {
@@ -534,6 +615,27 @@ class GraphvizGenerator : public TreeGenerator {
   }
 
  protected:
+  template <bool is_categorical>
+  std::string BuildEdge(RegTree const &tree, bst_node_t nid, int32_t child, bool left) const {
+    static std::string const kEdgeTemplate =
+        "    {nid} -> {child} [label=\"{branch}\" color=\"{color}\"]\n";
+    // Is this the default child for missing value?
+    bool is_missing = tree[nid].DefaultChild() == child;
+    std::string branch;
+    if (is_categorical) {
+      branch = std::string{left ? "no" : "yes"} + std::string{is_missing ? ", missing" : ""};
+    } else {
+      branch = std::string{left ? "yes" : "no"} + std::string{is_missing ? ", missing" : ""};
+    }
+    std::string buffer =
+        SuperT::Match(kEdgeTemplate,
+                {{"{nid}", std::to_string(nid)},
+                 {"{child}", std::to_string(child)},
+                 {"{color}", is_missing ? param_.yes_color : param_.no_color},
+                 {"{branch}", branch}});
+    return buffer;
+  }
+
   // Only indicator is different, so we combine all different node types into this
   // function.
   std::string PlainNode(RegTree const& tree, int32_t nid, uint32_t) const override {
@@ -552,27 +654,32 @@ class GraphvizGenerator : public TreeGenerator {
         {"{cond}",   has_less ? SuperT::ToStr(cond) : ""},
         {"{params}", param_.condition_node_params}});
 
-    static std::string const kEdgeTemplate =
+    result += BuildEdge<false>(tree, nid, tree[nid].LeftChild(), true);
-        "    {nid} -> {child} [label=\"{branch}\" color=\"{color}\"]\n";
+    result += BuildEdge<false>(tree, nid, tree[nid].RightChild(), false);
-    auto MatchFn = SuperT::Match;  // mingw failed to capture protected fn.
+
-    auto BuildEdge =
-        [&tree, nid, MatchFn, this](int32_t child, bool left) {
-          // Is this the default child for missing value?
-          bool is_missing = tree[nid].DefaultChild() == child;
-          std::string branch = std::string {left ? "yes" : "no"} +
-                               std::string {is_missing ? ", missing" : ""};
-          std::string buffer = MatchFn(kEdgeTemplate, {
-              {"{nid}",        std::to_string(nid)},
-              {"{child}",      std::to_string(child)},
-              {"{color}",      is_missing ? param_.yes_color : param_.no_color},
-              {"{branch}",     branch}});
-          return buffer;
-        };
-    result += BuildEdge(tree[nid].LeftChild(), true);
-    result += BuildEdge(tree[nid].RightChild(), false);
     return result;
   };
 
+  std::string Categorical(RegTree const& tree, int32_t nid, uint32_t) const override {
+    static std::string const kLabelTemplate =
+        "    {nid} [ label=\"{fname}:{cond}\" {params}]\n";
+    auto cats = GetSplitCategories(tree, nid);
+    auto cats_str = PrintCatsAsSet(cats);
+    auto split = tree[nid].SplitIndex();
+    std::string result = SuperT::Match(
+        kLabelTemplate,
+        {{"{nid}", std::to_string(nid)},
+         {"{fname}", split < fmap_.Size() ? fmap_.Name(split)
+                                          : 'f' + std::to_string(split)},
+         {"{cond}", cats_str},
+         {"{params}", param_.condition_node_params}});
+
+    result += BuildEdge<true>(tree, nid, tree[nid].LeftChild(), true);
+    result += BuildEdge<true>(tree, nid, tree[nid].RightChild(), false);
+
+    return result;
+  }
+
   std::string LeafNode(RegTree const& tree, int32_t nid, uint32_t) const override {
     static std::string const kLeafTemplate =
         "    {nid} [ label=\"leaf={leaf-value}\" {params}]\n";
@@ -588,9 +695,12 @@ class GraphvizGenerator : public TreeGenerator {
       return this->LeafNode(tree, nid, depth);
     }
     static std::string const kNodeTemplate = "{parent}\n{left}\n{right}";
+    auto node = tree.GetSplitTypes()[nid] == FeatureType::kCategorical
+                    ? this->Categorical(tree, nid, depth)
+                    : this->PlainNode(tree, nid, depth);
     auto result = SuperT::Match(
         kNodeTemplate,
-        {{"{parent}", this->PlainNode(tree, nid, depth)},
+        {{"{parent}", node},
          {"{left}",   this->BuildTree(tree, tree[nid].LeftChild(), depth+1)},
          {"{right}",  this->BuildTree(tree, tree[nid].RightChild(), depth+1)}});
     return result;

tests/cpp/tree/test_tree_model.cc

@@ -241,6 +241,65 @@ RegTree ConstructTree() {
       /*right_sum=*/0.0f);
   return tree;
 }
+
+RegTree ConstructTreeCat(std::vector<bst_cat_t>* cond) {
+  RegTree tree;
+  std::vector<uint32_t> cats_storage(common::CatBitField::ComputeStorageSize(33), 0);
+  common::CatBitField split_cats(cats_storage);
+  split_cats.Set(0);
+  split_cats.Set(14);
+  split_cats.Set(32);
+
+  cond->push_back(0);
+  cond->push_back(14);
+  cond->push_back(32);
+
+  tree.ExpandCategorical(0, /*split_index=*/0, cats_storage, true, 0.0f, 2.0,
+                         3.00, 11.0, 2.0, 3.0, 4.0);
+  auto left = tree[0].LeftChild();
+  auto right = tree[0].RightChild();
+  tree.ExpandNode(
+      /*nid=*/left, /*split_index=*/1, /*split_value=*/1.0f,
+      /*default_left=*/false, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, /*left_sum=*/0.0f,
+      /*right_sum=*/0.0f);
+  tree.ExpandCategorical(right, /*split_index=*/0, cats_storage, true, 0.0f,
+                         2.0, 3.00, 11.0, 2.0, 3.0, 4.0);
+  return tree;
+}
+
+void TestCategoricalTreeDump(std::string format, std::string sep) {
+  std::vector<bst_cat_t> cond;
+  auto tree = ConstructTreeCat(&cond);
+
+  FeatureMap fmap;
+  auto str = tree.DumpModel(fmap, true, format);
+  std::string cond_str;
+  for (size_t c = 0; c < cond.size(); ++c) {
+    cond_str += std::to_string(cond[c]);
+    if (c != cond.size() - 1) {
+      cond_str += sep;
+    }
+  }
+  auto pos = str.find(cond_str);
+  ASSERT_NE(pos, std::string::npos);
+  pos = str.find(cond_str, pos + 1);
+  ASSERT_NE(pos, std::string::npos);
+
+  fmap.PushBack(0, "feat_0", "categorical");
+  fmap.PushBack(1, "feat_1", "q");
+  fmap.PushBack(2, "feat_2", "int");
+
+  str = tree.DumpModel(fmap, true, format);
+  pos = str.find(cond_str);
+  ASSERT_NE(pos, std::string::npos);
+  pos = str.find(cond_str, pos + 1);
+  ASSERT_NE(pos, std::string::npos);
+
+  if (format == "json") {
+    // Make sure it's valid JSON
+    Json::Load(StringView{str});
+  }
+}
 }  // anonymous namespace
 
 TEST(Tree, DumpJson) {
@@ -278,6 +337,10 @@ TEST(Tree, DumpJson) {
   ASSERT_EQ(get<Array>(j_tree["children"]).size(), 2ul);
 }
 
+TEST(Tree, DumpJsonCategorical) {
+  TestCategoricalTreeDump("json", ", ");
+}
+
 TEST(Tree, DumpText) {
   auto tree = ConstructTree();
   FeatureMap fmap;
@@ -313,6 +376,10 @@ TEST(Tree, DumpText) {
   ASSERT_EQ(str.find("cover"), std::string::npos);
 }
 
+TEST(Tree, DumpTextCategorical) {
+  TestCategoricalTreeDump("text", ",");
+}
+
 TEST(Tree, DumpDot) {
   auto tree = ConstructTree();
   FeatureMap fmap;
@@ -350,6 +417,10 @@ TEST(Tree, DumpDot) {
   ASSERT_NE(str.find(R"(1 -> 4 [label="no, missing")"), std::string::npos);
 }
 
+TEST(Tree, DumpDotCategorical) {
+  TestCategoricalTreeDump("dot", ",");
+}
+
 TEST(Tree, JsonIO) {
   RegTree tree;
   tree.ExpandNode(0, 0, 0.0f, false, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f,

tests/python-gpu/test_gpu_plotting.py

@@ -0,0 +1,40 @@
+import sys
+import xgboost as xgb
+import pytest
+import json
+
+sys.path.append("tests/python")
+import testing as tm
+
+try:
+    import matplotlib
+
+    matplotlib.use("Agg")
+    from matplotlib.axes import Axes
+    from graphviz import Source
+except ImportError:
+    pass
+
+
+pytestmark = pytest.mark.skipif(**tm.no_multiple(tm.no_matplotlib(), tm.no_graphviz()))
+
+
+class TestPlotting:
+    @pytest.mark.skipif(**tm.no_pandas())
+    def test_categorical(self):
+        X, y = tm.make_categorical(1000, 31, 19, onehot=False)
+        reg = xgb.XGBRegressor(
+            enable_categorical=True, n_estimators=10, tree_method="gpu_hist"
+        )
+        reg.fit(X, y)
+        trees = reg.get_booster().get_dump(dump_format="json")
+        for tree in trees:
+            j_tree = json.loads(tree)
+            assert "leaf" in j_tree.keys() or isinstance(
+                j_tree["split_condition"], list
+            )
+
+        graph = xgb.to_graphviz(reg, num_trees=len(j_tree) - 1)
+        assert isinstance(graph, Source)
+        ax = xgb.plot_tree(reg, num_trees=len(j_tree) - 1)
+        assert isinstance(ax, Axes)

tests/python-gpu/test_gpu_updaters.py

@@ -71,7 +71,6 @@ class TestGPUUpdaters:
     @settings(deadline=None)
     @pytest.mark.skipif(**tm.no_pandas())
     def test_categorical(self, rows, cols, rounds, cats):
-        pytest.xfail(reason='TestGPUUpdaters::test_categorical is flaky')
         self.run_categorical_basic(rows, cols, rounds, cats)
 
     def test_categorical_32_cat(self):

tests/python-gpu/test_gpu_with_sklearn.py

@@ -55,7 +55,6 @@ def test_categorical():
         tree_method="gpu_hist",
         use_label_encoder=False,
         enable_categorical=True,
-        predictor="gpu_predictor",
         n_estimators=10,
     )
     X = pd.DataFrame(X.todense()).astype("category")