scikit-learn · thomasjpfan · Dec 13, 2019 · Dec 5, 2019 · Dec 5, 2019 · Dec 5, 2019
diff --git a/doc/modules/ensemble.rst b/doc/modules/ensemble.rst
@@ -1323,7 +1323,7 @@ computationally expensive.
     StackingRegressor(...)
     >>> print('R2 score: {:.2f}'
     ...       .format(multi_layer_regressor.score(X_test, y_test)))
-    R2 score: 0.82
+    R2 score: 0.83
 
 .. topic:: References
 

diff --git a/sklearn/decomposition/_nmf.py b/sklearn/decomposition/_nmf.py
@@ -842,7 +842,7 @@ def _fit_multiplicative_update(X, W, H, beta_loss='frobenius',
 
 
 def non_negative_factorization(X, W=None, H=None, n_components=None,
-                               init='warn', update_H=True, solver='cd',
+                               init=None, update_H=True, solver='cd',
                                beta_loss='frobenius', tol=1e-4,
                                max_iter=200, alpha=0., l1_ratio=0.,
                                regularization=None, random_state=None,
@@ -891,10 +891,7 @@ def non_negative_factorization(X, W=None, H=None, n_components=None,
 
     init : None | 'random' | 'nndsvd' | 'nndsvda' | 'nndsvdar' | 'custom'
         Method used to initialize the procedure.
-        Default: 'random'.
-
-        The default value will change from 'random' to None in version 0.23
-        to make it consistent with decomposition.NMF.
+        Default: None.
 
         Valid options:
 
@@ -915,6 +912,9 @@ def non_negative_factorization(X, W=None, H=None, n_components=None,
 
         - 'custom': use custom matrices W and H
 
+        .. versionchanged:: 0.23
+            The default value of `init` changed from 'random' to None in 0.23.
+
     update_H : boolean, default: True
         Set to True, both W and H will be estimated from initial guesses.
         Set to False, only W will be estimated.
@@ -1028,13 +1028,6 @@ def non_negative_factorization(X, W=None, H=None, n_components=None,
         raise ValueError("Tolerance for stopping criteria must be "
                          "positive; got (tol=%r)" % tol)
 
-    if init == "warn":
-        if n_components < n_features:
-            warnings.warn("The default value of init will change from "
-                          "random to None in 0.23 to make it consistent "
-                          "with decomposition.NMF.", FutureWarning)
-        init = "random"
-
     # check W and H, or initialize them
     if init == 'custom' and update_H:
         _check_init(H, (n_components, n_features), "NMF (input H)")

diff --git a/sklearn/decomposition/tests/test_nmf.py b/sklearn/decomposition/tests/test_nmf.py
@@ -224,10 +224,6 @@ def test_non_negative_factorization_checking():
     A = np.ones((2, 2))
     # Test parameters checking is public function
     nnmf = non_negative_factorization
-    msg = ("The default value of init will change from "
-           "random to None in 0.23 to make it consistent "
-           "with decomposition.NMF.")
-    assert_warns_message(FutureWarning, msg, nnmf, A, A, A, np.int64(1))
     msg = ("Number of components must be a positive integer; "
            "got (n_components=1.5)")
     assert_raise_message(ValueError, msg, nnmf, A, A, A, 1.5, 'random')

diff --git a/sklearn/discriminant_analysis.py b/sklearn/discriminant_analysis.py
@@ -423,7 +423,6 @@ def fit(self, X, y):
         y : array, shape (n_samples,)
             Target values.
         """
-        # FIXME: Future warning to be removed in 0.23
         X, y = check_X_y(X, y, ensure_min_samples=2, estimator=self,
                          dtype=[np.float64, np.float32])
         self.classes_ = unique_labels(y)
@@ -455,21 +454,11 @@ def fit(self, X, y):
             self._max_components = max_components
         else:
             if self.n_components > max_components:
-                warnings.warn(
+                raise ValueError(
                     "n_components cannot be larger than min(n_features, "
-                    "n_classes - 1). Using min(n_features, "
-                    "n_classes - 1) = min(%d, %d - 1) = %d components."
-                    % (X.shape[1], len(self.classes_), max_components),
-                    ChangedBehaviorWarning)
-                future_msg = ("In version 0.23, setting n_components > min("
-                              "n_features, n_classes - 1) will raise a "
-                              "ValueError. You should set n_components to None"
-                              " (default), or a value smaller or equal to "
-                              "min(n_features, n_classes - 1).")
-                warnings.warn(future_msg, FutureWarning)
-                self._max_components = max_components
-            else:
-                self._max_components = self.n_components
+                    "n_classes - 1)."
+                )
+            self._max_components = self.n_components
 
         if self.solver == 'svd':
             if self.shrinkage is not None:

diff --git a/sklearn/ensemble/_forest.py b/sklearn/ensemble/_forest.py
@@ -935,14 +935,14 @@ class RandomForestClassifier(ForestClassifier):
 
         .. versionadded:: 0.19
 
-    min_impurity_split : float, (default=1e-7)
+    min_impurity_split : float, (default=0)
         Threshold for early stopping in tree growth. A node will split
         if its impurity is above the threshold, otherwise it is a leaf.
 
         .. deprecated:: 0.19
            ``min_impurity_split`` has been deprecated in favor of
            ``min_impurity_decrease`` in 0.19. The default value of
-           ``min_impurity_split`` will change from 1e-7 to 0 in 0.23 and it
+           ``min_impurity_split`` has changed from 1e-7 to 0 in 0.23 and it
            will be removed in 0.25. Use ``min_impurity_decrease`` instead.
 
 
@@ -1253,14 +1253,14 @@ class RandomForestRegressor(ForestRegressor):
 
         .. versionadded:: 0.19
 
-    min_impurity_split : float, (default=1e-7)
+    min_impurity_split : float, (default=0)
         Threshold for early stopping in tree growth. A node will split
         if its impurity is above the threshold, otherwise it is a leaf.
 
         .. deprecated:: 0.19
            ``min_impurity_split`` has been deprecated in favor of
            ``min_impurity_decrease`` in 0.19. The default value of
-           ``min_impurity_split`` will change from 1e-7 to 0 in 0.23 and it
+           ``min_impurity_split`` has changed from 1e-7 to 0 in 0.23 and it
            will be removed in 0.25. Use ``min_impurity_decrease`` instead.
 
     bootstrap : boolean, optional (default=True)
@@ -1530,14 +1530,14 @@ class ExtraTreesClassifier(ForestClassifier):
 
         .. versionadded:: 0.19
 
-    min_impurity_split : float, (default=1e-7)
+    min_impurity_split : float, (default=0)
         Threshold for early stopping in tree growth. A node will split
         if its impurity is above the threshold, otherwise it is a leaf.
 
         .. deprecated:: 0.19
            ``min_impurity_split`` has been deprecated in favor of
            ``min_impurity_decrease`` in 0.19. The default value of
-           ``min_impurity_split`` will change from 1e-7 to 0 in 0.23 and it
+           ``min_impurity_split`` has changed from 1e-7 to 0 in 0.23 and it
            will be removed in 0.25. Use ``min_impurity_decrease`` instead.
 
     bootstrap : boolean, optional (default=False)
@@ -1840,14 +1840,14 @@ class ExtraTreesRegressor(ForestRegressor):
 
         .. versionadded:: 0.19
 
-    min_impurity_split : float, (default=1e-7)
+    min_impurity_split : float, (default=0)
         Threshold for early stopping in tree growth. A node will split
         if its impurity is above the threshold, otherwise it is a leaf.
 
         .. deprecated:: 0.19
            ``min_impurity_split`` has been deprecated in favor of
            ``min_impurity_decrease`` in 0.19. The default value of
-           ``min_impurity_split`` will change from 1e-7 to 0 in 0.23 and it
+           ``min_impurity_split`` has changed from 1e-7 to 0 in 0.23 and it
            will be removed in 0.25. Use ``min_impurity_decrease`` instead.
 
     bootstrap : boolean, optional (default=False)
@@ -2078,14 +2078,14 @@ class RandomTreesEmbedding(BaseForest):
 
         .. versionadded:: 0.19
 
-    min_impurity_split : float, (default=1e-7)
+    min_impurity_split : float, (default=0)
         Threshold for early stopping in tree growth. A node will split
         if its impurity is above the threshold, otherwise it is a leaf.
 
         .. deprecated:: 0.19
            ``min_impurity_split`` has been deprecated in favor of
            ``min_impurity_decrease`` in 0.19. The default value of
-           ``min_impurity_split`` will change from 1e-7 to 0 in 0.23 and it
+           ``min_impurity_split`` has changed from 1e-7 to 0 in 0.23 and it
            will be removed in 0.25. Use ``min_impurity_decrease`` instead.
 
     sparse_output : bool, optional (default=True)

diff --git a/sklearn/ensemble/_gb.py b/sklearn/ensemble/_gb.py
@@ -868,14 +868,14 @@ class GradientBoostingClassifier(ClassifierMixin, BaseGradientBoosting):
 
         .. versionadded:: 0.19
 
-    min_impurity_split : float, (default=1e-7)
+    min_impurity_split : float, (default=0)
         Threshold for early stopping in tree growth. A node will split
         if its impurity is above the threshold, otherwise it is a leaf.
 
         .. deprecated:: 0.19
            ``min_impurity_split`` has been deprecated in favor of
            ``min_impurity_decrease`` in 0.19. The default value of
-           ``min_impurity_split`` will change from 1e-7 to 0 in 0.23 and it
+           ``min_impurity_split`` has changed from 1e-7 to 0 in 0.23 and it
            will be removed in 0.25. Use ``min_impurity_decrease`` instead.
 
     init : estimator or 'zero', optional (default=None)
@@ -1340,14 +1340,14 @@ class GradientBoostingRegressor(RegressorMixin, BaseGradientBoosting):
 
         .. versionadded:: 0.19
 
-    min_impurity_split : float, (default=1e-7)
+    min_impurity_split : float, (default=0)
         Threshold for early stopping in tree growth. A node will split
         if its impurity is above the threshold, otherwise it is a leaf.
 
         .. deprecated:: 0.19
            ``min_impurity_split`` has been deprecated in favor of
            ``min_impurity_decrease`` in 0.19. The default value of
-           ``min_impurity_split`` will change from 1e-7 to 0 in 0.23 and it
+           ``min_impurity_split`` has changed from 1e-7 to 0 in 0.23 and it
            will be removed in 0.25. Use ``min_impurity_decrease`` instead.
 
     init : estimator or 'zero', optional (default=None)

diff --git a/sklearn/ensemble/tests/test_gradient_boosting.py b/sklearn/ensemble/tests/test_gradient_boosting.py
@@ -1170,9 +1170,10 @@ def test_non_uniform_weights_toy_edge_case_clf():
 
 def check_sparse_input(EstimatorClass, X, X_sparse, y):
     dense = EstimatorClass(n_estimators=10, random_state=0,
-                           max_depth=2).fit(X, y)
+                           max_depth=2, min_impurity_decrease=1e-7).fit(X, y)
     sparse = EstimatorClass(n_estimators=10, random_state=0,
-                            max_depth=2).fit(X_sparse, y)
+                            max_depth=2,
+                            min_impurity_decrease=1e-7).fit(X_sparse, y)
 
     assert_array_almost_equal(sparse.apply(X), dense.apply(X))
     assert_array_almost_equal(sparse.predict(X), dense.predict(X))

diff --git a/sklearn/metrics/_classification.py b/sklearn/metrics/_classification.py
@@ -1986,7 +1986,7 @@ class 2       1.00      0.67      0.80         3
         return report
 
 
-def hamming_loss(y_true, y_pred, labels=None, sample_weight=None):
+def hamming_loss(y_true, y_pred, sample_weight=None):
     """Compute the average Hamming loss.
 
     The Hamming loss is the fraction of labels that are incorrectly predicted.
@@ -2001,17 +2001,6 @@ def hamming_loss(y_true, y_pred, labels=None, sample_weight=None):
     y_pred : 1d array-like, or label indicator array / sparse matrix
         Predicted labels, as returned by a classifier.
 
-    labels : array, shape = [n_labels], optional (default='deprecated')
-        Integer array of labels. If not provided, labels will be inferred
-        from y_true and y_pred.
-
-        .. versionadded:: 0.18
-        .. deprecated:: 0.21
-           This parameter ``labels`` is deprecated in version 0.21 and will
-           be removed in version 0.23. Hamming loss uses ``y_true.shape[1]``
-           for the number of labels when y_true is binary label indicators,
-           so it is unnecessary for the user to specify.
-
     sample_weight : array-like of shape (n_samples,), default=None
         Sample weights.
 
@@ -2071,12 +2060,6 @@ def hamming_loss(y_true, y_pred, labels=None, sample_weight=None):
     y_type, y_true, y_pred = _check_targets(y_true, y_pred)
     check_consistent_length(y_true, y_pred, sample_weight)
 
-    if labels is not None:
-        warnings.warn("The labels parameter is unused. It was"
-                      " deprecated in version 0.21 and"
-                      " will be removed in version 0.23",
-                      FutureWarning)
-
     if sample_weight is None:
         weight_average = 1.
     else:

diff --git a/sklearn/metrics/tests/test_classification.py b/sklearn/metrics/tests/test_classification.py
@@ -1176,11 +1176,6 @@ def test_multilabel_hamming_loss():
     assert hamming_loss(y1, np.zeros_like(y1), sample_weight=w) == 2. / 3
     # sp_hamming only works with 1-D arrays
     assert hamming_loss(y1[0], y2[0]) == sp_hamming(y1[0], y2[0])
-    assert_warns_message(FutureWarning,
-                         "The labels parameter is unused. It was"
-                         " deprecated in version 0.21 and"
-                         " will be removed in version 0.23",
-                         hamming_loss, y1, y2, labels=[0, 1])
 
 
 def test_jaccard_score_validation():

diff --git a/sklearn/metrics/tests/test_common.py b/sklearn/metrics/tests/test_common.py
@@ -351,8 +351,6 @@ def precision_recall_curve_padded_thresholds(*args, **kwargs):
     "roc_curve",
     "precision_recall_curve",
 
-    "hamming_loss",
-
     "precision_score", "recall_score", "f1_score", "f2_score", "f0.5_score",
     "jaccard_score",
 

diff --git a/sklearn/preprocessing/_data.py b/sklearn/preprocessing/_data.py
@@ -2606,8 +2606,8 @@ def quantile_transform(X, axis=0, n_quantiles=1000,
         input is already a numpy array). If True, a copy of `X` is transformed,
         leaving the original `X` unchanged
 
-        ..versionchnanged:: 0.22
-            The default value of `copy` changed from False to True in 0.22.
+        ..versionchnanged:: 0.23
+            The default value of `copy` changed from False to True in 0.23.
 
     Returns
     -------
@@ -3008,7 +3008,7 @@ def _more_tags(self):
         return {'allow_nan': True}
 
 
-def power_transform(X, method='warn', standardize=True, copy=True):
+def power_transform(X, method='yeo-johnson', standardize=True, copy=True):
     """
     Power transforms are a family of parametric, monotonic transformations
     that are applied to make data more Gaussian-like. This is useful for
@@ -3032,15 +3032,15 @@ def power_transform(X, method='warn', standardize=True, copy=True):
     X : array-like, shape (n_samples, n_features)
         The data to be transformed using a power transformation.
 
-    method : str
+    method : {'yeo-johnson', 'box-cox'}, default='yeo-johnson'
         The power transform method. Available methods are:
 
         - 'yeo-johnson' [1]_, works with positive and negative values
         - 'box-cox' [2]_, only works with strictly positive values
 
-        The default method will be changed from 'box-cox' to 'yeo-johnson'
-        in version 0.23. To suppress the FutureWarning, explicitly set the
-        parameter.
+        .. versionchanged:: 0.23
+            The default value of the `method` parameter changed from
+            'box-cox' to 'yeo-johnson' in 0.23.
 
     standardize : boolean, default=True
         Set to True to apply zero-mean, unit-variance normalization to the
@@ -3092,12 +3092,5 @@ def power_transform(X, method='warn', standardize=True, copy=True):
     .. [2] G.E.P. Box and D.R. Cox, "An Analysis of Transformations", Journal
            of the Royal Statistical Society B, 26, 211-252 (1964).
     """
-    if method == 'warn':
-        warnings.warn("The default value of 'method' will change from "
-                      "'box-cox' to 'yeo-johnson' in version 0.23. Set "
-                      "the 'method' argument explicitly to silence this "
-                      "warning in the meantime.",
-                      FutureWarning)
-        method = 'box-cox'
     pt = PowerTransformer(method=method, standardize=standardize, copy=copy)
     return pt.fit_transform(X)
diff --git a/sklearn/preprocessing/tests/test_data.py b/sklearn/preprocessing/tests/test_data.py
@@ -2452,21 +2452,3 @@ def test_power_transformer_copy_False(method, standardize):
 
     X_inv_trans = pt.inverse_transform(X_trans)
     assert X_trans is X_inv_trans
-
-
-def test_power_transform_default_method():
-    X = np.abs(X_2d)
-
-    future_warning_message = (
-        "The default value of 'method' "
-        "will change from 'box-cox'"
-    )
-    assert_warns_message(FutureWarning, future_warning_message,
-                         power_transform, X)
-
-    with warnings.catch_warnings():
-        warnings.simplefilter('ignore')
-        X_trans_default = power_transform(X)
-
-    X_trans_boxcox = power_transform(X, method='box-cox')
-    assert_array_equal(X_trans_boxcox, X_trans_default)