fix #865

src/boosting/gbdt.cpp

@@ -115,7 +115,7 @@ void GBDT::Init(const BoostingConfig* config, const Dataset* train_data, const O
   feature_infos_ = train_data_->feature_infos();
 
   // if need bagging, create buffer
-  ResetBaggingConfig(gbdt_config_.get());
+  ResetBaggingConfig(gbdt_config_.get(), true);
 
   // reset config for tree learner
   class_need_train_ = std::vector<bool>(num_tree_per_iteration_, true);
@@ -211,7 +211,7 @@ void GBDT::ResetTrainingData(const Dataset* train_data, const ObjectiveFunction*
 
     feature_infos_ = train_data_->feature_infos();
 
-    ResetBaggingConfig(gbdt_config_.get());
+    ResetBaggingConfig(gbdt_config_.get(), true);
 
     tree_learner_->ResetTrainingData(train_data);
   }
@@ -222,13 +222,13 @@ void GBDT::ResetConfig(const BoostingConfig* config) {
   early_stopping_round_ = new_config->early_stopping_round;
   shrinkage_rate_ = new_config->learning_rate;
   if (tree_learner_ != nullptr) {
-    ResetBaggingConfig(new_config.get());
+    ResetBaggingConfig(new_config.get(), false);
     tree_learner_->ResetConfig(&new_config->tree_config);
   }
   gbdt_config_.reset(new_config.release());
 }
 
-void GBDT::ResetBaggingConfig(const BoostingConfig* config) {
+void GBDT::ResetBaggingConfig(const BoostingConfig* config, bool is_change_dataset) {
   // if need bagging, create buffer
   if (config->bagging_fraction < 1.0 && config->bagging_freq > 0) {
     bag_data_cnt_ =
@@ -252,8 +252,10 @@ void GBDT::ResetBaggingConfig(const BoostingConfig* config) {
     const int sparse_group_threshold_usesubset = train_data_->num_feature_groups() / 4;
     if (average_bag_rate <= 0.5
         && (train_data_->num_feature_groups() < group_threshold_usesubset || sparse_group < sparse_group_threshold_usesubset)) {
-      tmp_subset_.reset(new Dataset(bag_data_cnt_));
-      tmp_subset_->CopyFeatureMapperFrom(train_data_);
+      if (tmp_subset_ == nullptr || is_change_dataset) {
+        tmp_subset_.reset(new Dataset(bag_data_cnt_));
+        tmp_subset_->CopyFeatureMapperFrom(train_data_);
+      }
       is_use_subset_ = true;
       Log::Debug("use subset for bagging");
     }
@@ -263,6 +265,10 @@ void GBDT::ResetBaggingConfig(const BoostingConfig* config) {
     tmp_indices_.clear();
     is_use_subset_ = false;
   }
+
+  if (is_change_dataset) {
+    need_re_bagging_ = true;
+  }
 }
 
 void GBDT::AddValidDataset(const Dataset* valid_data,
@@ -322,7 +328,8 @@ data_size_t GBDT::BaggingHelper(Random& cur_rand, data_size_t start, data_size_t
 
 void GBDT::Bagging(int iter) {
   // if need bagging
-  if (bag_data_cnt_ < num_data_ && iter % gbdt_config_->bagging_freq == 0) {
+  if ( (bag_data_cnt_ < num_data_ && iter % gbdt_config_->bagging_freq == 0) || need_re_bagging_) {
+    need_re_bagging_ = false;
     const data_size_t min_inner_size = 1000;
     data_size_t inner_size = (num_data_ + num_threads_ - 1) / num_threads_;
     if (inner_size < min_inner_size) { inner_size = min_inner_size; }

src/boosting/gbdt.h

@@ -270,7 +270,7 @@ public:
   virtual const char* SubModelName() const override { return "tree"; }
 
 protected:
-  void ResetBaggingConfig(const BoostingConfig* config);
+  void ResetBaggingConfig(const BoostingConfig* config, bool is_change_dataset);
   /*!
   * \brief Implement bagging logic
   * \param iter Current interation
@@ -388,6 +388,7 @@ protected:
   bool is_constant_hessian_;
   std::unique_ptr<ObjectiveFunction> loaded_objective_;
   bool average_output_;
+  bool need_re_bagging_ = false;
 };
 
 }  // namespace LightGBM