fix param aliases (#4387)

R-package/R/aliases.R

@@ -4,7 +4,7 @@
 # [description] List of respected parameter aliases specific to lgb.Dataset. Wrapped in a function to
 #               take advantage of lazy evaluation (so it doesn't matter what order
 #               R sources files during installation).
-# [return] A named list, where each key is a parameter relevant to lgb.DataSet and each value is a character
+# [return] A named list, where each key is a parameter relevant to lgb.Dataset and each value is a character
 #          vector of corresponding aliases.
 .DATASET_PARAMETERS <- function() {
     return(
@@ -57,6 +57,10 @@
                 "label_column"
                 , "label"
             )
+            , "linear_tree" = c(
+                "linear_tree"
+                , "linear_trees"
+            )
             , "max_bin" = "max_bin"
             , "max_bin_by_feature" = "max_bin_by_feature"
             , "min_data_in_bin" = "min_data_in_bin"
@@ -64,6 +68,7 @@
                 "pre_partition"
                 , "is_pre_partition"
             )
+            , "precise_float_parser" = "precise_float_parser"
             , "two_round" = c(
                 "two_round"
                 , "two_round_loading"

docs/Parameters.rst

@@ -139,28 +139,6 @@ Core Parameters
 
       -  **Note**: internally, LightGBM uses ``gbdt`` mode for the first ``1 / learning_rate`` iterations
 
--  ``linear_tree`` :raw-html:`<a id="linear_tree" title="Permalink to this parameter" href="#linear_tree">&#x1F517;&#xFE0E;</a>`, default = ``false``, type = bool, aliases: ``linear_trees``
-
-   -  fit piecewise linear gradient boosting tree
-
-      -  tree splits are chosen in the usual way, but the model at each leaf is linear instead of constant
-
-      -  the linear model at each leaf includes all the numerical features in that leaf's branch
-
-      -  categorical features are used for splits as normal but are not used in the linear models
-
-      -  missing values should not be encoded as ``0``. Use ``np.nan`` for Python, ``NA`` for the CLI, and ``NA``, ``NA_real_``, or ``NA_integer_`` for R
-
-      -  it is recommended to rescale data before training so that features have similar mean and standard deviation
-
-      -  **Note**: only works with CPU and ``serial`` tree learner
-
-      -  **Note**: ``regression_l1`` objective is not supported with linear tree boosting
-
-      -  **Note**: setting ``linear_tree=true`` significantly increases the memory use of LightGBM
-
-      -  **Note**: if you specify ``monotone_constraints``, constraints will be enforced when choosing the split points, but not when fitting the linear models on leaves
-
 -  ``data`` :raw-html:`<a id="data" title="Permalink to this parameter" href="#data">&#x1F517;&#xFE0E;</a>`, default = ``""``, type = string, aliases: ``train``, ``train_data``, ``train_data_file``, ``data_filename``
 
    -  path of training data, LightGBM will train from this data
@@ -672,6 +650,28 @@ IO Parameters
 Dataset Parameters
 ~~~~~~~~~~~~~~~~~~
 
+-  ``linear_tree`` :raw-html:`<a id="linear_tree" title="Permalink to this parameter" href="#linear_tree">&#x1F517;&#xFE0E;</a>`, default = ``false``, type = bool, aliases: ``linear_trees``
+
+   -  fit piecewise linear gradient boosting tree
+
+      -  tree splits are chosen in the usual way, but the model at each leaf is linear instead of constant
+
+      -  the linear model at each leaf includes all the numerical features in that leaf's branch
+
+      -  categorical features are used for splits as normal but are not used in the linear models
+
+      -  missing values should not be encoded as ``0``. Use ``np.nan`` for Python, ``NA`` for the CLI, and ``NA``, ``NA_real_``, or ``NA_integer_`` for R
+
+      -  it is recommended to rescale data before training so that features have similar mean and standard deviation
+
+      -  **Note**: only works with CPU and ``serial`` tree learner
+
+      -  **Note**: ``regression_l1`` objective is not supported with linear tree boosting
+
+      -  **Note**: setting ``linear_tree=true`` significantly increases the memory use of LightGBM
+
+      -  **Note**: if you specify ``monotone_constraints``, constraints will be enforced when choosing the split points, but not when fitting the linear models on leaves
+
 -  ``max_bin`` :raw-html:`<a id="max_bin" title="Permalink to this parameter" href="#max_bin">&#x1F517;&#xFE0E;</a>`, default = ``255``, type = int, constraints: ``max_bin > 1``
 
    -  max number of bins that feature values will be bucketed in

include/LightGBM/config.h

@@ -149,19 +149,6 @@ struct Config {
   // descl2 = **Note**: internally, LightGBM uses ``gbdt`` mode for the first ``1 / learning_rate`` iterations
   std::string boosting = "gbdt";
 
-  // alias = linear_trees
-  // desc = fit piecewise linear gradient boosting tree
-  // descl2 = tree splits are chosen in the usual way, but the model at each leaf is linear instead of constant
-  // descl2 = the linear model at each leaf includes all the numerical features in that leaf's branch
-  // descl2 = categorical features are used for splits as normal but are not used in the linear models
-  // descl2 = missing values should not be encoded as ``0``. Use ``np.nan`` for Python, ``NA`` for the CLI, and ``NA``, ``NA_real_``, or ``NA_integer_`` for R
-  // descl2 = it is recommended to rescale data before training so that features have similar mean and standard deviation
-  // descl2 = **Note**: only works with CPU and ``serial`` tree learner
-  // descl2 = **Note**: ``regression_l1`` objective is not supported with linear tree boosting
-  // descl2 = **Note**: setting ``linear_tree=true`` significantly increases the memory use of LightGBM
-  // descl2 = **Note**: if you specify ``monotone_constraints``, constraints will be enforced when choosing the split points, but not when fitting the linear models on leaves
-  bool linear_tree = false;
-
   // alias = train, train_data, train_data_file, data_filename
   // desc = path of training data, LightGBM will train from this data
   // desc = **Note**: can be used only in CLI version
@@ -586,6 +573,19 @@ struct Config {
 
   #pragma region Dataset Parameters
 
+  // alias = linear_trees
+  // desc = fit piecewise linear gradient boosting tree
+  // descl2 = tree splits are chosen in the usual way, but the model at each leaf is linear instead of constant
+  // descl2 = the linear model at each leaf includes all the numerical features in that leaf's branch
+  // descl2 = categorical features are used for splits as normal but are not used in the linear models
+  // descl2 = missing values should not be encoded as ``0``. Use ``np.nan`` for Python, ``NA`` for the CLI, and ``NA``, ``NA_real_``, or ``NA_integer_`` for R
+  // descl2 = it is recommended to rescale data before training so that features have similar mean and standard deviation
+  // descl2 = **Note**: only works with CPU and ``serial`` tree learner
+  // descl2 = **Note**: ``regression_l1`` objective is not supported with linear tree boosting
+  // descl2 = **Note**: setting ``linear_tree=true`` significantly increases the memory use of LightGBM
+  // descl2 = **Note**: if you specify ``monotone_constraints``, constraints will be enforced when choosing the split points, but not when fitting the linear models on leaves
+  bool linear_tree = false;
+
   // check = >1
   // desc = max number of bins that feature values will be bucketed in
   // desc = small number of bins may reduce training accuracy but may increase general power (deal with over-fitting)

python-package/lightgbm/basic.py

@@ -311,6 +311,8 @@ class _ConfigAliases:
                                     "sparse"},
                "label_column": {"label_column",
                                 "label"},
+               "linear_tree": {"linear_tree",
+                               "linear_trees"},
                "local_listen_port": {"local_listen_port",
                                      "local_port",
                                      "port"},
@@ -1144,6 +1146,7 @@ class Dataset:
                                                 "max_bin_by_feature",
                                                 "min_data_in_bin",
                                                 "pre_partition",
+                                                "precise_float_parser",
                                                 "two_round",
                                                 "use_missing",
                                                 "weight_column",
@@ -3180,7 +3183,11 @@ class Booster:
         _safe_call(_LIB.LGBM_BoosterGetLinear(
             self.handle,
             ctypes.byref(out_is_linear)))
-        new_params = deepcopy(self.params)
+        new_params = _choose_param_value(
+            main_param_name="linear_tree",
+            params=self.params,
+            default_value=None
+        )
         new_params["linear_tree"] = out_is_linear.value
         train_set = Dataset(data, label, silent=True, params=new_params)
         new_params['refit_decay_rate'] = decay_rate

src/c_api.cpp

@@ -287,9 +287,13 @@ class Booster {
           "You need to set `feature_pre_filter=false` to dynamically change "
           "the `min_data_in_leaf`.");
     }
-    if (new_param.count("linear_tree") && (new_config.linear_tree != old_config.linear_tree)) {
+    if (new_param.count("linear_tree") && new_config.linear_tree != old_config.linear_tree) {
       Log::Fatal("Cannot change linear_tree after constructed Dataset handle.");
     }
+    if (new_param.count("precise_float_parser") &&
+        new_config.precise_float_parser != old_config.precise_float_parser) {
+      Log::Fatal("Cannot change precise_float_parser after constructed Dataset handle.");
+    }
   }
 
   void ResetConfig(const char* parameters) {

src/io/config_auto.cpp

@@ -16,7 +16,6 @@ const std::unordered_map<std::string, std::string>& Config::alias_table() {
   {"application", "objective"},
   {"boosting_type", "boosting"},
   {"boost", "boosting"},
-  {"linear_trees", "linear_tree"},
   {"train", "data"},
   {"train_data", "data"},
   {"train_data_file", "data"},
@@ -106,6 +105,7 @@ const std::unordered_map<std::string, std::string>& Config::alias_table() {
   {"model_output", "output_model"},
   {"model_out", "output_model"},
   {"save_period", "snapshot_freq"},
+  {"linear_trees", "linear_tree"},
   {"subsample_for_bin", "bin_construct_sample_cnt"},
   {"data_seed", "data_random_seed"},
   {"is_sparse", "is_enable_sparse"},
@@ -176,7 +176,6 @@ const std::unordered_set<std::string>& Config::parameter_set() {
   "task",
   "objective",
   "boosting",
-  "linear_tree",
   "data",
   "valid",
   "num_iterations",
@@ -241,6 +240,7 @@ const std::unordered_set<std::string>& Config::parameter_set() {
   "output_model",
   "saved_feature_importance_type",
   "snapshot_freq",
+  "linear_tree",
   "max_bin",
   "max_bin_by_feature",
   "min_data_in_bin",
@@ -309,8 +309,6 @@ const std::unordered_set<std::string>& Config::parameter_set() {
 
 void Config::GetMembersFromString(const std::unordered_map<std::string, std::string>& params) {
   std::string tmp_str = "";
-  GetBool(params, "linear_tree", &linear_tree);
-
   GetString(params, "data", &data);
 
   if (GetString(params, "valid", &tmp_str)) {
@@ -483,6 +481,8 @@ void Config::GetMembersFromString(const std::unordered_map<std::string, std::str
 
   GetInt(params, "snapshot_freq", &snapshot_freq);
 
+  GetBool(params, "linear_tree", &linear_tree);
+
   GetInt(params, "max_bin", &max_bin);
   CHECK_GT(max_bin, 1);
 
@@ -634,7 +634,6 @@ void Config::GetMembersFromString(const std::unordered_map<std::string, std::str
 
 std::string Config::SaveMembersToString() const {
   std::stringstream str_buf;
-  str_buf << "[linear_tree: " << linear_tree << "]\n";
   str_buf << "[data: " << data << "]\n";
   str_buf << "[valid: " << Common::Join(valid, ",") << "]\n";
   str_buf << "[num_iterations: " << num_iterations << "]\n";
@@ -693,6 +692,7 @@ std::string Config::SaveMembersToString() const {
   str_buf << "[interaction_constraints: " << interaction_constraints << "]\n";
   str_buf << "[verbosity: " << verbosity << "]\n";
   str_buf << "[saved_feature_importance_type: " << saved_feature_importance_type << "]\n";
+  str_buf << "[linear_tree: " << linear_tree << "]\n";
   str_buf << "[max_bin: " << max_bin << "]\n";
   str_buf << "[max_bin_by_feature: " << Common::Join(max_bin_by_feature, ",") << "]\n";
   str_buf << "[min_data_in_bin: " << min_data_in_bin << "]\n";

tests/python_package_test/test_engine.py

@@ -2345,6 +2345,7 @@ def test_dataset_update_params():
                       "ignore_column": 0,
                       "min_data_in_leaf": 10,
                       "linear_tree": False,
+                      "precise_float_parser": True,
                       "verbose": -1}
     unchangeable_params = {"max_bin": 150,
                            "max_bin_by_feature": [30, 5],
@@ -2366,7 +2367,8 @@ def test_dataset_update_params():
                            "ignore_column": 1,
                            "forcedbins_filename": "/some/path/forcedbins.json",
                            "min_data_in_leaf": 2,
-                           "linear_tree": True}
+                           "linear_tree": True,
+                           "precise_float_parser": False}
     X = np.random.random((100, 2))
     y = np.random.random(100)