Always use partition based categorical splits.

doc/tutorials/categorical.rst

@@ -72,23 +72,20 @@ Optimal Partitioning
 .. versionadded:: 1.6
 
 Optimal partitioning is a technique for partitioning the categorical predictors for each
-node split, the proof of optimality for numerical objectives like ``RMSE`` was first
-introduced by `[1] <#references>`__. The algorithm is used in decision trees for handling
-regression and binary classification tasks `[2] <#references>`__, later LightGBM `[3]
-<#references>`__ brought it to the context of gradient boosting trees and now is also
-adopted in XGBoost as an optional feature for handling categorical splits. More
-specifically, the proof by Fisher `[1] <#references>`__ states that, when trying to
-partition a set of discrete values into groups based on the distances between a measure of
-these values, one only needs to look at sorted partitions instead of enumerating all
-possible permutations. In the context of decision trees, the discrete values are
-categories, and the measure is the output leaf value.  Intuitively, we want to group the
-categories that output similar leaf values. During split finding, we first sort the
-gradient histogram to prepare the contiguous partitions then enumerate the splits
+node split, the proof of optimality for numerical output was first introduced by `[1]
+<#references>`__. The algorithm is used in decision trees `[2] <#references>`__, later
+LightGBM `[3] <#references>`__ brought it to the context of gradient boosting trees and
+now is also adopted in XGBoost as an optional feature for handling categorical
+splits. More specifically, the proof by Fisher `[1] <#references>`__ states that, when
+trying to partition a set of discrete values into groups based on the distances between a
+measure of these values, one only needs to look at sorted partitions instead of
+enumerating all possible permutations. In the context of decision trees, the discrete
+values are categories, and the measure is the output leaf value.  Intuitively, we want to
+group the categories that output similar leaf values. During split finding, we first sort
+the gradient histogram to prepare the contiguous partitions then enumerate the splits
 according to these sorted values. One of the related parameters for XGBoost is
 ``max_cat_to_one_hot``, which controls whether one-hot encoding or partitioning should be
-used for each feature, see :doc:`/parameter` for details.  When objective is not
-regression or binary classification, XGBoost will fallback to using onehot encoding
-instead.
+used for each feature, see :doc:`/parameter` for details.
 
 
 **********************

include/xgboost/task.h

@@ -38,9 +38,6 @@ struct ObjInfo {
   ObjInfo(Task t) : task{t} {}  // NOLINT
   ObjInfo(Task t, bool khess, bool zhess) : task{t}, const_hess{khess}, zero_hess(zhess) {}
 
-  XGBOOST_DEVICE bool UseOneHot() const {
-    return (task != ObjInfo::kRegression && task != ObjInfo::kBinary);
-  }
   /**
    * \brief Use adaptive tree if the objective doesn't have valid hessian value.
    */

src/common/categorical.h

@@ -83,7 +83,7 @@ inline void InvalidCategory() {
  * \brief Whether should we use onehot encoding for categorical data.
  */
 XGBOOST_DEVICE inline bool UseOneHot(uint32_t n_cats, uint32_t max_cat_to_onehot, ObjInfo task) {
-  bool use_one_hot = n_cats < max_cat_to_onehot || task.UseOneHot();
+  bool use_one_hot = n_cats < max_cat_to_onehot;
   return use_one_hot;
 }
 

src/tree/gpu_hist/evaluator.cu

@@ -21,7 +21,7 @@ void GPUHistEvaluator<GradientSumT>::Reset(common::HistogramCuts const &cuts,
                                            int32_t device) {
   param_ = param;
   tree_evaluator_ = TreeEvaluator{param, n_features, device};
-  if (cuts.HasCategorical() && !task.UseOneHot()) {
+  if (cuts.HasCategorical()) {
     dh::XGBCachingDeviceAllocator<char> alloc;
     auto ptrs = cuts.cut_ptrs_.ConstDeviceSpan();
     auto beg = thrust::make_counting_iterator<size_t>(1ul);