Enable parameter validation for skl.

python-package/xgboost/core.py

@@ -1098,6 +1098,7 @@ def _init_from_array_interface(self, data, missing, nthread):
                 ctypes.c_int(self.max_bin), ctypes.byref(handle)))
         self.handle = handle
 
+
 class Booster(object):
     # pylint: disable=too-many-public-methods
     """A Booster of XGBoost.
@@ -1129,10 +1130,12 @@ def __init__(self, params=None, cache=(), model_file=None):
         self.handle = ctypes.c_void_p()
         _check_call(_LIB.XGBoosterCreate(dmats, c_bst_ulong(len(cache)),
                                          ctypes.byref(self.handle)))
+        params = params or {}
+        if isinstance(params, list):
+            params.append(('validate_parameters', True))
+        else:
+            params['validate_parameters'] = True
 
-        if isinstance(params, dict) and \
-            'validate_parameters' not in params.keys():
-            params['validate_parameters'] = 1
         self.set_param(params or {})
         if (params is not None) and ('booster' in params):
             self.booster = params['booster']

python-package/xgboost/sklearn.py

@@ -210,7 +210,7 @@ def __init__(self, max_depth=None, learning_rate=None, n_estimators=100,
                  missing=np.nan, num_parallel_tree=None,
                  monotone_constraints=None, interaction_constraints=None,
                  importance_type="gain", gpu_id=None,
-                 validate_parameters=False, **kwargs):
+                 validate_parameters=None, **kwargs):
         if not SKLEARN_INSTALLED:
             raise XGBoostError(
                 'sklearn needs to be installed in order to use this module')
@@ -242,9 +242,6 @@ def __init__(self, max_depth=None, learning_rate=None, n_estimators=100,
         self.interaction_constraints = interaction_constraints
         self.importance_type = importance_type
         self.gpu_id = gpu_id
-        # Parameter validation is not working with Scikit-Learn interface, as
-        # it passes all paraemters into XGBoost core, whether they are used or
-        # not.
         self.validate_parameters = validate_parameters
 
     def get_booster(self):
@@ -340,9 +337,16 @@ def parse_parameter(value):
         return params
 
     def get_xgb_params(self):
-        """Get xgboost type parameters."""
-        xgb_params = self.get_params()
-        return xgb_params
+        """Get xgboost specific parameters."""
+        params = self.get_params()
+        # Parameters that should not go into native learner.
+        wrapper_specific = {
+            'importance_type', 'kwargs', 'missing', 'n_estimators'}
+        filtered = dict()
+        for k, v in params.items():
+            if k not in wrapper_specific:
+                filtered[k] = v
+        return filtered
 
     def get_num_boosting_rounds(self):
         """Gets the number of xgboost boosting rounds."""
@@ -540,7 +544,8 @@ def fit(self, X, y, sample_weight=None, base_margin=None,
         if evals_result:
             for val in evals_result.items():
                 evals_result_key = list(val[1].keys())[0]
-                evals_result[val[0]][evals_result_key] = val[1][evals_result_key]
+                evals_result[val[0]][evals_result_key] = val[1][
+                    evals_result_key]
             self.evals_result_ = evals_result
 
         if early_stopping_rounds is not None:

tests/python/test_with_sklearn.py

@@ -9,6 +9,8 @@
 import unittest
 import json
 
+from test_basic import captured_output
+
 rng = np.random.RandomState(1994)
 
 pytestmark = pytest.mark.skipif(**tm.no_sklearn())
@@ -265,7 +267,7 @@ def test_parameter_tuning():
     xgb_model = xgb.XGBRegressor(learning_rate=0.1)
     clf = GridSearchCV(xgb_model, {'max_depth': [2, 4, 6],
                                    'n_estimators': [50, 100, 200]},
-                       cv=3, verbose=1, iid=True)
+                       cv=3, verbose=1)
     clf.fit(X, y)
     assert clf.best_score_ < 0.7
     assert clf.best_params_ == {'n_estimators': 100, 'max_depth': 4}
@@ -785,6 +787,27 @@ def test_constraint_parameters():
         'train_param']['interaction_constraints'] == '[[0, 1], [2, 3, 4]]'
 
 
+def test_parameter_validation():
+    reg = xgb.XGBRegressor(foo='bar', verbosity=1)
+    X = np.random.randn(10, 10)
+    y = np.random.randn(10)
+    with captured_output() as (out, err):
+        reg.fit(X, y)
+        output = out.getvalue().strip()
+
+    assert output.find('foo') != -1
+
+    reg = xgb.XGBRegressor(n_estimators=2, missing=3,
+                           importance_type='gain', verbosity=1)
+    X = np.random.randn(10, 10)
+    y = np.random.randn(10)
+    with captured_output() as (out, err):
+        reg.fit(X, y)
+        output = out.getvalue().strip()
+
+    assert len(output) == 0
+
+
 class TestBoostFromPrediction(unittest.TestCase):
     def run_boost_from_prediction(self, tree_method):
         from sklearn.datasets import load_breast_cancer