do not modify users params (#1722)

python-package/lightgbm/basic.py

@@ -13,10 +13,11 @@ from tempfile import NamedTemporaryFile
 import numpy as np
 import scipy.sparse
 
-from .compat import (DataFrame, LGBMDeprecationWarning, Series,
-                     decode_string, integer_types,
+from .compat import (DataFrame, Series,
+                     decode_string, string_type,
+                     integer_types, numeric_types,
                      json, json_default_with_numpy,
-                     numeric_types, range_, zip_, string_type)
+                     range_, zip_)
 from .libpath import find_lib_path
 
 
@@ -1470,7 +1471,7 @@ class Booster(object):
         self.__set_objective_to_none = False
         self.best_iteration = -1
         self.best_score = {}
-        params = {} if params is None else params
+        params = {} if params is None else copy.deepcopy(params)
         # user can set verbose with params, it has higher priority
         if not any(verbose_alias in params for verbose_alias in ('verbose', 'verbosity')) and silent:
             params["verbose"] = -1
@@ -1540,7 +1541,7 @@ class Booster(object):
             self.model_from_string(params['model_str'])
         else:
             raise TypeError('Need at least one training dataset or model file to create booster instance')
-        self.params = params.copy()
+        self.params = params
 
     def __del__(self):
         try:
@@ -2139,7 +2140,7 @@ class Booster(object):
         result : numpy array
             Prediction result.
         """
-        predictor = self._to_predictor(kwargs)
+        predictor = self._to_predictor(copy.deepcopy(kwargs))
         if num_iteration is None:
             num_iteration = self.best_iteration
         return predictor.predict(data, num_iteration,
@@ -2169,7 +2170,7 @@ class Booster(object):
         """
         if self.__set_objective_to_none:
             raise LightGBMError('Cannot refit due to null objective function.')
-        predictor = self._to_predictor(kwargs)
+        predictor = self._to_predictor(copy.deepcopy(kwargs))
         leaf_preds = predictor.predict(data, -1, pred_leaf=True)
         nrow, ncol = leaf_preds.shape
         train_set = Dataset(data, label, silent=True)

python-package/lightgbm/engine.py

@@ -4,6 +4,7 @@
 from __future__ import absolute_import
 
 import collections
+import copy
 import warnings
 from operator import attrgetter
 
@@ -12,7 +13,7 @@ import numpy as np
 from . import callback
 from .basic import Booster, Dataset, LightGBMError, _InnerPredictor
 from .compat import (SKLEARN_INSTALLED, _LGBMGroupKFold, _LGBMStratifiedKFold,
-                     integer_types, range_, zip_, string_type)
+                     string_type, integer_types, range_, zip_)
 
 
 def train(params, train_set, num_boost_round=100,
@@ -104,6 +105,7 @@ def train(params, train_set, num_boost_round=100,
         The trained Booster model.
     """
     # create predictor first
+    params = copy.deepcopy(params)
     for alias in ["num_iterations", "num_iteration", "n_iter", "num_tree", "num_trees",
                   "num_round", "num_rounds", "num_boost_round", "n_estimators"]:
         if alias in params:
@@ -417,6 +419,7 @@ def cv(params, train_set, num_boost_round=100,
     if not isinstance(train_set, Dataset):
         raise TypeError("Traninig only accepts Dataset object")
 
+    params = copy.deepcopy(params)
     for alias in ["num_iterations", "num_iteration", "n_iter", "num_tree", "num_trees",
                   "num_round", "num_rounds", "num_boost_round", "n_estimators"]:
         if alias in params:

python-package/lightgbm/sklearn.py

@@ -11,7 +11,7 @@ from .compat import (SKLEARN_INSTALLED, _LGBMClassifierBase,
                      LGBMNotFittedError, _LGBMLabelEncoder, _LGBMModelBase,
                      _LGBMRegressorBase, _LGBMCheckXY, _LGBMCheckArray, _LGBMCheckConsistentLength,
                      _LGBMAssertAllFinite, _LGBMCheckClassificationTargets, _LGBMComputeSampleWeight,
-                     argc_, range_, string_type, DataFrame, LGBMDeprecationWarning)
+                     argc_, range_, string_type, DataFrame)
 from .engine import train
 
 

tests/python_package_test/test_engine.py

@@ -438,7 +438,8 @@ class TestEngine(unittest.TestCase):
         lgb_train = lgb.Dataset(X_train, y_train)
         # shuffle = False, override metric in params
         params_with_metric = {'metric': 'l2', 'verbose': -1}
-        cv_res = lgb.cv(params_with_metric, lgb_train, num_boost_round=10, nfold=3, stratified=False, shuffle=False,
+        cv_res = lgb.cv(params_with_metric, lgb_train, num_boost_round=10,
+                        nfold=3, stratified=False, shuffle=False,
                         metrics='l1', verbose_eval=False)
         self.assertIn('l1-mean', cv_res)
         self.assertNotIn('l2-mean', cv_res)
@@ -452,10 +453,10 @@ class TestEngine(unittest.TestCase):
         # self defined folds
         tss = TimeSeriesSplit(3)
         folds = tss.split(X_train)
-        cv_res_gen = lgb.cv(params, lgb_train, num_boost_round=10, folds=folds,
-                            metrics='l2', verbose_eval=False)
-        cv_res_obj = lgb.cv(params, lgb_train, num_boost_round=10, folds=tss,
-                            metrics='l2', verbose_eval=False)
+        cv_res_gen = lgb.cv(params_with_metric, lgb_train, num_boost_round=10, folds=folds,
+                            verbose_eval=False)
+        cv_res_obj = lgb.cv(params_with_metric, lgb_train, num_boost_round=10, folds=tss,
+                            verbose_eval=False)
         np.testing.assert_almost_equal(cv_res_gen['l2-mean'], cv_res_obj['l2-mean'])
         # lambdarank
         X_train, y_train = load_svmlight_file(os.path.join(os.path.dirname(os.path.realpath(__file__)),
@@ -464,21 +465,21 @@ class TestEngine(unittest.TestCase):
                                           '../../examples/lambdarank/rank.train.query'))
         params_lambdarank = {'objective': 'lambdarank', 'verbose': -1, 'eval_at': 3}
         lgb_train = lgb.Dataset(X_train, y_train, group=q_train)
-        # ... with NDCG (default) metric
-        cv_res_lambda = lgb.cv(params_lambdarank, lgb_train, num_boost_round=10, nfold=3,
-                               verbose_eval=False)
-        self.assertEqual(len(cv_res_lambda), 2)
-        self.assertFalse(np.isnan(cv_res_lambda['ndcg@3-mean']).any())
         # ... with l2 metric
         cv_res_lambda = lgb.cv(params_lambdarank, lgb_train, num_boost_round=10, nfold=3,
                                metrics='l2', verbose_eval=False)
         self.assertEqual(len(cv_res_lambda), 2)
         self.assertFalse(np.isnan(cv_res_lambda['l2-mean']).any())
+        # ... with NDCG (default) metric
+        cv_res_lambda = lgb.cv(params_lambdarank, lgb_train, num_boost_round=10, nfold=3,
+                               verbose_eval=False)
+        self.assertEqual(len(cv_res_lambda), 2)
+        self.assertFalse(np.isnan(cv_res_lambda['ndcg@3-mean']).any())
         # self defined folds with lambdarank
         cv_res_lambda_obj = lgb.cv(params_lambdarank, lgb_train, num_boost_round=10,
                                    folds=GroupKFold(n_splits=3),
-                                   metrics='l2', verbose_eval=False)
-        np.testing.assert_almost_equal(cv_res_lambda['l2-mean'], cv_res_lambda_obj['l2-mean'])
+                                   verbose_eval=False)
+        np.testing.assert_almost_equal(cv_res_lambda['ndcg@3-mean'], cv_res_lambda_obj['ndcg@3-mean'])
 
     def test_feature_name(self):
         X, y = load_boston(True)