Make QuantileDMatrix default to sklearn esitmators.

.github/workflows/r_tests.yml

@@ -31,8 +31,8 @@ jobs:
       uses: actions/cache@v2
       with:
         path: ${{ env.R_LIBS_USER }}
-        key: ${{ runner.os }}-r-${{ matrix.config.r }}-4-${{ hashFiles('R-package/DESCRIPTION') }}
-        restore-keys: ${{ runner.os }}-r-${{ matrix.config.r }}-4-${{ hashFiles('R-package/DESCRIPTION') }}
+        key: ${{ runner.os }}-r-${{ matrix.config.r }}-5-${{ hashFiles('R-package/DESCRIPTION') }}
+        restore-keys: ${{ runner.os }}-r-${{ matrix.config.r }}-5-${{ hashFiles('R-package/DESCRIPTION') }}
 
     - name: Install dependencies
       shell: Rscript {0}
@@ -80,20 +80,25 @@ jobs:
       uses: actions/cache@v2
       with:
         path: ${{ env.R_LIBS_USER }}
-        key: ${{ runner.os }}-r-${{ matrix.config.r }}-4-${{ hashFiles('R-package/DESCRIPTION') }}
-        restore-keys: ${{ runner.os }}-r-${{ matrix.config.r }}-4-${{ hashFiles('R-package/DESCRIPTION') }}
+        key: ${{ runner.os }}-r-${{ matrix.config.r }}-5-${{ hashFiles('R-package/DESCRIPTION') }}
+        restore-keys: ${{ runner.os }}-r-${{ matrix.config.r }}-5-${{ hashFiles('R-package/DESCRIPTION') }}
 
     - name: Install dependencies
       shell: Rscript {0}
+      if: matrix.config.os != 'windows-latest'
       run: |
         install.packages(${{ env.R_PACKAGES }},
                          repos = 'http://cloud.r-project.org',
                          dependencies = c('Depends', 'Imports', 'LinkingTo'))
-    - name: Install igraph on Windows
+
+    - name: Install binary dependencies
       shell: Rscript {0}
       if: matrix.config.os == 'windows-latest'
       run: |
-        install.packages('igraph', type='binary', dependencies = c('Depends', 'Imports', 'LinkingTo'))
+        install.packages(${{ env.R_PACKAGES }},
+                         type = 'binary',
+                         repos = 'http://cloud.r-project.org',
+                         dependencies = c('Depends', 'Imports', 'LinkingTo'))
 
     - uses: actions/setup-python@v2
       with:
@@ -132,8 +137,8 @@ jobs:
       uses: actions/cache@v2
       with:
         path: ${{ env.R_LIBS_USER }}
-        key: ${{ runner.os }}-r-${{ matrix.config.r }}-4-${{ hashFiles('R-package/DESCRIPTION') }}
-        restore-keys: ${{ runner.os }}-r-${{ matrix.config.r }}-4-${{ hashFiles('R-package/DESCRIPTION') }}
+        key: ${{ runner.os }}-r-${{ matrix.config.r }}-5-${{ hashFiles('R-package/DESCRIPTION') }}
+        restore-keys: ${{ runner.os }}-r-${{ matrix.config.r }}-5-${{ hashFiles('R-package/DESCRIPTION') }}
 
     - name: Install dependencies
       shell: Rscript {0}

.gitignore

@@ -97,8 +97,11 @@ metastore_db
 R-package/src/Makevars
 *.lib
 
-# Visual Studio Code
-/.vscode/
+# Visual Studio
+.vs/
+CMakeSettings.json
+*.ilk
+*.pdb
 
 # IntelliJ/CLion
 .idea

.readthedocs.yaml

@@ -0,0 +1,28 @@
+# .readthedocs.yaml
+# Read the Docs configuration file
+# See https://docs.readthedocs.io/en/stable/config-file/v2.html for details
+
+# Required
+version: 2
+
+# Set the version of Python and other tools you might need
+build:
+  os: ubuntu-22.04
+  tools:
+    python: "3.8"
+  apt_packages:
+    - graphviz
+
+# Build documentation in the docs/ directory with Sphinx
+sphinx:
+   configuration: doc/conf.py
+
+# If using Sphinx, optionally build your docs in additional formats such as PDF
+formats:
+   - pdf
+
+# Optionally declare the Python requirements required to build your docs
+python:
+  install:
+   - requirements: doc/requirements.txt
+  system_packages: true

R-package/src/Makevars.win

@@ -30,7 +30,7 @@ $(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_CXXFLAGS) $(SHLIB_PTHREAD_FLAGS)
-PKG_LIBS = $(SHLIB_OPENMP_CXXFLAGS) $(SHLIB_PTHREAD_FLAGS)
+PKG_LIBS = $(SHLIB_OPENMP_CXXFLAGS) $(SHLIB_PTHREAD_FLAGS) -lwsock32 -lws2_32
 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.o $(PKGROOT)/rabit/src/rabit_c_api.o \

cmake/Utils.cmake

@@ -244,7 +244,7 @@ macro(xgboost_target_properties target)
       $<$<NOT:$<COMPILE_LANGUAGE:CUDA>>:/utf-8>
       -D_CRT_SECURE_NO_WARNINGS
       -D_CRT_SECURE_NO_DEPRECATE
-      )
+    )
   endif (MSVC)
 
   if (WIN32 AND MINGW)
@@ -314,4 +314,8 @@ macro(xgboost_target_link_libraries target)
   if (RABIT_BUILD_MPI)
     target_link_libraries(${target} PRIVATE MPI::MPI_CXX)
   endif (RABIT_BUILD_MPI)
+
+  if (MINGW)
+    target_link_libraries(${target} PRIVATE wsock32 ws2_32)
+  endif (MINGW)
 endmacro(xgboost_target_link_libraries)

python-package/xgboost/dask.py

@@ -726,10 +726,9 @@ def _create_quantile_dmatrix(
     if parts is None:
         msg = f"worker {worker.address} has an empty DMatrix."
         LOGGER.warning(msg)
-        import cupy
 
         d = QuantileDMatrix(
-            cupy.zeros((0, 0)),
+            numpy.empty((0, 0)),
             feature_names=feature_names,
             feature_types=feature_types,
             max_bin=max_bin,
@@ -1544,15 +1543,21 @@ def inplace_predict(  # pylint: disable=unused-argument
 
 
 async def _async_wrap_evaluation_matrices(
-    client: Optional["distributed.Client"], **kwargs: Any
+    client: Optional["distributed.Client"],
+    tree_method: Optional[str],
+    max_bin: Optional[int],
+    **kwargs: Any,
 ) -> Tuple[DaskDMatrix, Optional[List[Tuple[DaskDMatrix, str]]]]:
     """A switch function for async environment."""
 
-    def _inner(**kwargs: Any) -> DaskDMatrix:
-        m = DaskDMatrix(client=client, **kwargs)
-        return m
+    def _dispatch(ref: Optional[DaskDMatrix], **kwargs: Any) -> DaskDMatrix:
+        if tree_method in ("hist", "gpu_hist"):
+            return DaskQuantileDMatrix(
+                client=client, ref=ref, max_bin=max_bin, **kwargs
+            )
+        return DaskDMatrix(client=client, **kwargs)
 
-    train_dmatrix, evals = _wrap_evaluation_matrices(create_dmatrix=_inner, **kwargs)
+    train_dmatrix, evals = _wrap_evaluation_matrices(create_dmatrix=_dispatch, **kwargs)
     train_dmatrix = await train_dmatrix
     if evals is None:
         return train_dmatrix, evals
@@ -1756,6 +1761,8 @@ async def _fit_async(
         params = self.get_xgb_params()
         dtrain, evals = await _async_wrap_evaluation_matrices(
             client=self.client,
+            tree_method=self.tree_method,
+            max_bin=self.max_bin,
             X=X,
             y=y,
             group=None,
@@ -1851,6 +1858,8 @@ async def _fit_async(
         params = self.get_xgb_params()
         dtrain, evals = await _async_wrap_evaluation_matrices(
             self.client,
+            tree_method=self.tree_method,
+            max_bin=self.max_bin,
             X=X,
             y=y,
             group=None,
@@ -2057,6 +2066,8 @@ async def _fit_async(
         params = self.get_xgb_params()
         dtrain, evals = await _async_wrap_evaluation_matrices(
             self.client,
+            tree_method=self.tree_method,
+            max_bin=self.max_bin,
             X=X,
             y=y,
             group=None,

python-package/xgboost/sklearn.py

@@ -38,6 +38,7 @@
     Booster,
     DMatrix,
     Metric,
+    QuantileDMatrix,
     XGBoostError,
     _convert_ntree_limit,
     _deprecate_positional_args,
@@ -430,7 +431,8 @@ def _wrap_evaluation_matrices(
     enable_categorical: bool,
     feature_types: Optional[FeatureTypes],
 ) -> Tuple[Any, List[Tuple[Any, str]]]:
-    """Convert array_like evaluation matrices into DMatrix.  Perform validation on the way."""
+    """Convert array_like evaluation matrices into DMatrix.  Perform validation on the
+    way."""
     train_dmatrix = create_dmatrix(
         data=X,
         label=y,
@@ -442,6 +444,7 @@ def _wrap_evaluation_matrices(
         missing=missing,
         enable_categorical=enable_categorical,
         feature_types=feature_types,
+        ref=None,
     )
 
     n_validation = 0 if eval_set is None else len(eval_set)
@@ -491,6 +494,7 @@ def validate_or_none(meta: Optional[Sequence], name: str) -> Sequence:
                     missing=missing,
                     enable_categorical=enable_categorical,
                     feature_types=feature_types,
+                    ref=train_dmatrix,
                 )
                 evals.append(m)
         nevals = len(evals)
@@ -904,6 +908,17 @@ def _duplicated(parameter: str) -> None:
 
         return model, metric, params, early_stopping_rounds, callbacks
 
+    def _create_dmatrix(self, ref: Optional[DMatrix], **kwargs: Any) -> DMatrix:
+        # Use `QuantileDMatrix` to save memory.
+        if self.tree_method in ("hist", "gpu_hist"):
+            try:
+                return QuantileDMatrix(
+                    **kwargs, ref=ref, nthread=self.n_jobs, max_bin=self.max_bin
+                )
+            except TypeError:  # `QuantileDMatrix` supports lesser types than DMatrix
+                pass
+        return DMatrix(**kwargs, nthread=self.n_jobs)
+
     def _set_evaluation_result(self, evals_result: TrainingCallback.EvalsLog) -> None:
         if evals_result:
             self.evals_result_ = cast(Dict[str, Dict[str, List[float]]], evals_result)
@@ -996,7 +1011,7 @@ def fit(
                 base_margin_eval_set=base_margin_eval_set,
                 eval_group=None,
                 eval_qid=None,
-                create_dmatrix=lambda **kwargs: DMatrix(nthread=self.n_jobs, **kwargs),
+                create_dmatrix=self._create_dmatrix,
                 enable_categorical=self.enable_categorical,
                 feature_types=self.feature_types,
             )
@@ -1479,7 +1494,7 @@ def fit(
                 base_margin_eval_set=base_margin_eval_set,
                 eval_group=None,
                 eval_qid=None,
-                create_dmatrix=lambda **kwargs: DMatrix(nthread=self.n_jobs, **kwargs),
+                create_dmatrix=self._create_dmatrix,
                 enable_categorical=self.enable_categorical,
                 feature_types=self.feature_types,
             )
@@ -1930,7 +1945,7 @@ def fit(
                 base_margin_eval_set=base_margin_eval_set,
                 eval_group=eval_group,
                 eval_qid=eval_qid,
-                create_dmatrix=lambda **kwargs: DMatrix(nthread=self.n_jobs, **kwargs),
+                create_dmatrix=self._create_dmatrix,
                 enable_categorical=self.enable_categorical,
                 feature_types=self.feature_types,
             )

rabit/src/allreduce_base.cc

@@ -5,7 +5,10 @@
  *
  * \author Tianqi Chen, Ignacio Cano, Tianyi Zhou
  */
+#if !defined(NOMINMAX) && defined(_WIN32)
 #define NOMINMAX
+#endif  // !defined(NOMINMAX)
+
 #include "rabit/base.h"
 #include "rabit/internal/rabit-inl.h"
 #include "allreduce_base.h"

src/cli_main.cc

@@ -6,7 +6,11 @@
  */
 #define _CRT_SECURE_NO_WARNINGS
 #define _CRT_SECURE_NO_DEPRECATE
+
+#if !defined(NOMINMAX) && defined(_WIN32)
 #define NOMINMAX
+#endif  // !defined(NOMINMAX)
+
 #include <dmlc/timer.h>
 
 #include <xgboost/learner.h>

src/data/iterative_dmatrix.cc

@@ -7,13 +7,46 @@
 
 #include "../common/column_matrix.h"
 #include "../common/hist_util.h"
+#include "../tree/param.h"  // FIXME(jiamingy): Find a better way to share this parameter.
 #include "gradient_index.h"
 #include "proxy_dmatrix.h"
 #include "simple_batch_iterator.h"
 
 namespace xgboost {
 namespace data {
 
+IterativeDMatrix::IterativeDMatrix(DataIterHandle iter_handle, DMatrixHandle proxy,
+                                   std::shared_ptr<DMatrix> ref, DataIterResetCallback* reset,
+                                   XGDMatrixCallbackNext* next, float missing, int nthread,
+                                   bst_bin_t max_bin)
+    : proxy_{proxy}, reset_{reset}, next_{next} {
+  // fetch the first batch
+  auto iter =
+      DataIterProxy<DataIterResetCallback, XGDMatrixCallbackNext>{iter_handle, reset_, next_};
+  iter.Reset();
+  bool valid = iter.Next();
+  CHECK(valid) << "Iterative DMatrix must have at least 1 batch.";
+
+  auto d = MakeProxy(proxy_)->DeviceIdx();
+
+  StringView msg{"All batch should be on the same device."};
+  if (batch_param_.gpu_id != Context::kCpuId) {
+    CHECK_EQ(d, batch_param_.gpu_id) << msg;
+  }
+
+  batch_param_ = BatchParam{d, max_bin};
+  // hardcoded parameter.
+  batch_param_.sparse_thresh = tree::TrainParam::DftSparseThreshold();
+
+  ctx_.UpdateAllowUnknown(
+      Args{{"nthread", std::to_string(nthread)}, {"gpu_id", std::to_string(d)}});
+  if (ctx_.IsCPU()) {
+    this->InitFromCPU(iter_handle, missing, ref);
+  } else {
+    this->InitFromCUDA(iter_handle, missing, ref);
+  }
+}
+
 void GetCutsFromRef(std::shared_ptr<DMatrix> ref_, bst_feature_t n_features, BatchParam p,
                     common::HistogramCuts* p_cuts) {
   CHECK(ref_);
@@ -199,6 +232,7 @@ void IterativeDMatrix::InitFromCPU(DataIterHandle iter_handle, float missing,
   if (n_batches == 1) {
     this->info_ = std::move(proxy->Info());
     this->info_.num_nonzero_ = nnz;
+    this->info_.num_col_ = n_features;  // proxy might be empty.
     CHECK_EQ(proxy->Info().labels.Size(), 0);
   }
 }
@@ -210,6 +244,10 @@ BatchSet<GHistIndexMatrix> IterativeDMatrix::GetGradientIndex(BatchParam const&
     ghist_ = std::make_shared<GHistIndexMatrix>(&ctx_, Info(), *ellpack_, param);
   }
 
+  if (param.sparse_thresh != tree::TrainParam::DftSparseThreshold()) {
+    LOG(WARNING) << "`sparse_threshold` can not be changed when `QuantileDMatrix` is used instead "
+                    "of `DMatrix`.";
+  }
   auto begin_iter =
       BatchIterator<GHistIndexMatrix>(new SimpleBatchIteratorImpl<GHistIndexMatrix>(ghist_));
   return BatchSet<GHistIndexMatrix>(begin_iter);

src/data/iterative_dmatrix.cu

@@ -173,8 +173,15 @@ BatchSet<EllpackPage> IterativeDMatrix::GetEllpackBatches(BatchParam const& para
   }
   if (!ellpack_ && ghist_) {
     ellpack_.reset(new EllpackPage());
-    this->ctx_.gpu_id = param.gpu_id;
-    this->Info().feature_types.SetDevice(param.gpu_id);
+    // Evaluation QuantileDMatrix initialized from CPU data might not have the correct GPU
+    // ID.
+    if (this->ctx_.IsCPU()) {
+      this->ctx_.gpu_id = param.gpu_id;
+    }
+    if (this->ctx_.IsCPU()) {
+      this->ctx_.gpu_id = dh::CurrentDevice();
+    }
+    this->Info().feature_types.SetDevice(this->ctx_.gpu_id);
     *ellpack_->Impl() =
         EllpackPageImpl(&ctx_, *this->ghist_, this->Info().feature_types.ConstDeviceSpan());
   }