MNT deprecate attributes in Partial Least Squares module

doc/whats_new/v0.24.rst

@@ -149,6 +149,13 @@ Changelog
   retrieved by calling `transform` on the training data. The `norm_y_weights`
   attribute will also be removed. :pr:`17095` by `Nicolas Hug`_.
 
+- |API| For :class:`cross_decomposition.PLSRegression`,
+  :class:`cross_decomposition.PLSCanonical`,
+  :class:`cross_decomposition.CCA`, and
+  :class:`cross_decomposition.PLSSVD`, the `x_mean_`, `y_mean_`, `x_std_`, and
+  `y_std_` attributes were deprecated and will be removed in 0.26.
+  :pr:`18768` by :user:`Maren Westermann <marenwestermann>`.
+
 :mod:`sklearn.datasets`
 .......................
 

sklearn/cross_decomposition/_pls.py

@@ -215,7 +215,7 @@ def fit(self, X, Y):
         norm_y_weights = self._norm_y_weights
 
         # Scale (in place)
-        Xk, Yk, self.x_mean_, self.y_mean_, self.x_std_, self.y_std_ = (
+        Xk, Yk, self._x_mean, self._y_mean, self._x_std, self._y_std = (
             _center_scale_xy(X, Y, self.scale))
 
         self.x_weights_ = np.zeros((p, n_components))  # U
@@ -294,7 +294,7 @@ def fit(self, X, Y):
                                    check_finite=False))
 
         self.coef_ = np.dot(self.x_rotations_, self.y_loadings_.T)
-        self.coef_ = self.coef_ * self.y_std_
+        self.coef_ = self.coef_ * self._y_std
         return self
 
     def transform(self, X, Y=None, copy=True):
@@ -318,16 +318,16 @@ def transform(self, X, Y=None, copy=True):
         check_is_fitted(self)
         X = check_array(X, copy=copy, dtype=FLOAT_DTYPES)
         # Normalize
-        X -= self.x_mean_
-        X /= self.x_std_
+        X -= self._x_mean
+        X /= self._x_std
         # Apply rotation
         x_scores = np.dot(X, self.x_rotations_)
         if Y is not None:
             Y = check_array(Y, ensure_2d=False, copy=copy, dtype=FLOAT_DTYPES)
             if Y.ndim == 1:
                 Y = Y.reshape(-1, 1)
-            Y -= self.y_mean_
-            Y /= self.y_std_
+            Y -= self._y_mean
+            Y /= self._y_std
             y_scores = np.dot(Y, self.y_rotations_)
             return x_scores, y_scores
 
@@ -356,8 +356,8 @@ def inverse_transform(self, X):
         X_reconstructed = np.matmul(X, self.x_loadings_.T)
 
         # Denormalize
-        X_reconstructed *= self.x_std_
-        X_reconstructed += self.x_mean_
+        X_reconstructed *= self._x_std
+        X_reconstructed += self._x_mean
         return X_reconstructed
 
     def predict(self, X, copy=True):
@@ -380,10 +380,10 @@ def predict(self, X, copy=True):
         check_is_fitted(self)
         X = check_array(X, copy=copy, dtype=FLOAT_DTYPES)
         # Normalize
-        X -= self.x_mean_
-        X /= self.x_std_
+        X -= self._x_mean
+        X /= self._x_std
         Ypred = np.dot(X, self.coef_)
-        return Ypred + self.y_mean_
+        return Ypred + self._y_mean
 
     def fit_transform(self, X, y=None):
         """Learn and apply the dimension reduction on the train data.
@@ -412,6 +412,34 @@ def fit_transform(self, X, y=None):
     def norm_y_weights(self):
         return self._norm_y_weights
 
+    @deprecated(  # type: ignore
+        "Attribute x_mean_ was deprecated in version 0.24 and "
+        "will be removed in 0.26.")
+    @property
+    def x_mean_(self):
+        return self._x_mean
+
+    @deprecated(  # type: ignore
+        "Attribute y_mean_ was deprecated in version 0.24 and "
+        "will be removed in 0.26.")
+    @property
+    def y_mean_(self):
+        return self._y_mean
+
+    @deprecated(  # type: ignore
+        "Attribute x_std_ was deprecated in version 0.24 and "
+        "will be removed in 0.26.")
+    @property
+    def x_std_(self):
+        return self._x_std
+
+    @deprecated(  # type: ignore
+        "Attribute y_std_ was deprecated in version 0.24 and "
+        "will be removed in 0.26.")
+    @property
+    def y_std_(self):
+        return self._y_std
+
     @property
     def x_scores_(self):
         # TODO: raise error in 0.26 instead
@@ -870,7 +898,7 @@ def fit(self, X, Y):
             )
             n_components = rank_upper_bound
 
-        X, Y, self.x_mean_, self.y_mean_, self.x_std_, self.y_std_ = (
+        X, Y, self._x_mean, self._y_mean, self._x_std, self._y_std = (
             _center_scale_xy(X, Y, self.scale))
 
         # Compute SVD of cross-covariance matrix
@@ -905,6 +933,34 @@ def x_scores_(self):
     def y_scores_(self):
         return self._y_scores
 
+    @deprecated(  # type: ignore
+        "Attribute x_mean_ was deprecated in version 0.24 and "
+        "will be removed in 0.26.")
+    @property
+    def x_mean_(self):
+        return self._x_mean
+
+    @deprecated(  # type: ignore
+        "Attribute y_mean_ was deprecated in version 0.24 and "
+        "will be removed in 0.26.")
+    @property
+    def y_mean_(self):
+        return self._y_mean
+
+    @deprecated(  # type: ignore
+        "Attribute x_std_ was deprecated in version 0.24 and "
+        "will be removed in 0.26.")
+    @property
+    def x_std_(self):
+        return self._x_std
+
+    @deprecated(  # type: ignore
+        "Attribute y_std_ was deprecated in version 0.24 and "
+        "will be removed in 0.26.")
+    @property
+    def y_std_(self):
+        return self._y_std
+
     def transform(self, X, Y=None):
         """
         Apply the dimensionality reduction.
@@ -926,13 +982,13 @@ def transform(self, X, Y=None):
         """
         check_is_fitted(self)
         X = check_array(X, dtype=np.float64)
-        Xr = (X - self.x_mean_) / self.x_std_
+        Xr = (X - self._x_mean) / self._x_std
         x_scores = np.dot(Xr, self.x_weights_)
         if Y is not None:
             Y = check_array(Y, ensure_2d=False, dtype=np.float64)
             if Y.ndim == 1:
                 Y = Y.reshape(-1, 1)
-            Yr = (Y - self.y_mean_) / self.y_std_
+            Yr = (Y - self._y_mean) / self._y_std
             y_scores = np.dot(Yr, self.y_weights_)
             return x_scores, y_scores
         return x_scores

sklearn/cross_decomposition/tests/test_pls.py

@@ -488,6 +488,20 @@ def test_norm_y_weights_deprecation(Est):
         est.norm_y_weights
 
 
+# TODO: Remove test in 0.26
+@pytest.mark.parametrize('Estimator',
+                         (PLSRegression, PLSCanonical, CCA, PLSSVD))
+@pytest.mark.parametrize('attribute',
+                         ("x_mean_", "y_mean_", "x_std_", "y_std_"))
+def test_mean_and_std_deprecation(Estimator, attribute):
+    rng = np.random.RandomState(0)
+    X = rng.randn(10, 5)
+    Y = rng.randn(10, 3)
+    estimator = Estimator().fit(X, Y)
+    with pytest.warns(FutureWarning, match=f"{attribute} was deprecated"):
+        getattr(estimator, attribute)
+
+
 @pytest.mark.parametrize('n_samples, n_features', [(100, 10), (100, 200)])
 @pytest.mark.parametrize('seed', range(10))
 def test_singular_value_helpers(n_samples, n_features, seed):

sklearn/tests/test_docstring_parameters.py

@@ -183,13 +183,12 @@ def test_fit_docstring_attributes(name, Estimator):
     doc = docscrape.ClassDoc(Estimator)
     attributes = doc['Attributes']
 
-    IGNORED = {'CCA', 'ClassifierChain', 'ColumnTransformer',
+    IGNORED = {'ClassifierChain', 'ColumnTransformer',
                'CountVectorizer', 'DictVectorizer', 'FeatureUnion',
                'GaussianRandomProjection',
                'MultiOutputClassifier', 'MultiOutputRegressor',
                'NoSampleWeightWrapper', 'OneVsOneClassifier',
-               'OutputCodeClassifier', 'Pipeline', 'PLSCanonical',
-               'PLSRegression', 'PLSSVD', 'RFE', 'RFECV',
+               'OutputCodeClassifier', 'Pipeline', 'RFE', 'RFECV',
                'RegressorChain', 'SelectFromModel',
                'SparseCoder', 'SparseRandomProjection',
                'SpectralBiclustering', 'StackingClassifier',
@@ -252,10 +251,8 @@ def test_fit_docstring_attributes(name, Estimator):
         with ignore_warnings(category=FutureWarning):
             assert hasattr(est, attr.name)
 
-    IGNORED = {'Birch', 'CCA',
-               'LarsCV', 'Lasso',
-               'OrthogonalMatchingPursuit',
-               'PLSCanonical', 'PLSSVD'}
+    IGNORED = {'Birch', 'LarsCV', 'Lasso',
+               'OrthogonalMatchingPursuit'}
 
     if Estimator.__name__ in IGNORED:
         pytest.xfail(