add sklearn_tags

feature_engine/_base_transformers/base_numerical.py

@@ -22,7 +22,7 @@
 
 
 class BaseNumericalTransformer(
-    BaseEstimator, TransformerMixin, GetFeatureNamesOutMixin
+    TransformerMixin, BaseEstimator, GetFeatureNamesOutMixin
 ):
     """Shared set-up procedures across numerical transformers, i.e.,
     variable transformers, discretisers, math combination.
@@ -122,3 +122,7 @@ def _more_tags(self):
         tags_dict = _return_tags()
         tags_dict["variables"] = "numerical"
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        return tags

feature_engine/_prediction/base_predictor.py

@@ -298,3 +298,7 @@ def _predict(self, X: pd.DataFrame) -> np.ndarray:
 
     def _more_tags(self):
         return _return_tags()
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        return tags

feature_engine/_prediction/target_mean_classifier.py

@@ -6,7 +6,7 @@
 from feature_engine._prediction.base_predictor import BaseTargetMeanEstimator
 
 
-class TargetMeanClassifier(BaseTargetMeanEstimator, ClassifierMixin):
+class TargetMeanClassifier(ClassifierMixin, BaseTargetMeanEstimator):
     """
     The TargetMeanClassifier() estimates target values based on the average of the mean
     target value per category or bin of a group of categorical and numerical variables.
@@ -185,3 +185,8 @@ def predict(self, X: pd.DataFrame) -> np.ndarray:
         """
         y_pred = np.where(self._predict(X) > 0.5, self.classes_[1], self.classes_[0])
         return y_pred
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        tags.classifier_tags.multi_class = False
+        return tags

feature_engine/_prediction/target_mean_regressor.py

@@ -6,7 +6,7 @@
 from feature_engine._prediction.base_predictor import BaseTargetMeanEstimator
 
 
-class TargetMeanRegressor(BaseTargetMeanEstimator, RegressorMixin):
+class TargetMeanRegressor(RegressorMixin, BaseTargetMeanEstimator):
     """
     The TargetMeanRegressor() outputs a target estimation based on the mean target
     value per category or bin, across a group of categorical or numerical variables.

feature_engine/creation/base_creation.py

@@ -22,7 +22,7 @@
 )
 
 
-class BaseCreation(BaseEstimator, TransformerMixin, GetFeatureNamesOutMixin):
+class BaseCreation(TransformerMixin, BaseEstimator, GetFeatureNamesOutMixin):
     """Shared set-up, checks and methods across creation transformers."""
 
     def __init__(
@@ -128,3 +128,8 @@ def _more_tags(self):
         ] = "this transformer works with datasets that contain at least 2 variables. \
         Otherwise, there is nothing to combine"
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        tags.input_tags.allow_nan = True
+        return tags

feature_engine/creation/decision_tree_features.py

@@ -58,7 +58,7 @@
     transform=_transform_creation_docstring,
     fit_transform=_fit_transform_docstring,
 )
-class DecisionTreeFeatures(BaseEstimator, TransformerMixin, GetFeatureNamesOutMixin):
+class DecisionTreeFeatures(TransformerMixin, BaseEstimator, GetFeatureNamesOutMixin):
     """
     `DecisionTreeFeatures()` adds new variables to the data that result of the output of
     decision trees trained with one or more features.
@@ -471,3 +471,7 @@ def _more_tags(self):
         tags_dict["requires_y"] = True
         tags_dict["variables"] = "numerical"
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        return tags

feature_engine/datetime/datetime.py

@@ -45,7 +45,7 @@
     fit=_fit_not_learn_docstring,
     fit_transform=_fit_transform_docstring,
 )
-class DatetimeFeatures(BaseEstimator, TransformerMixin, GetFeatureNamesOutMixin):
+class DatetimeFeatures(TransformerMixin, BaseEstimator, GetFeatureNamesOutMixin):
     """
     DatetimeFeatures extracts date and time features from datetime variables, adding
     new columns to the dataset. DatetimeFeatures can extract datetime information from
@@ -394,3 +394,7 @@ def _check_index_contains_na(self, index: pd.Index):
     def _more_tags(self):
         tags_dict = {"variables": "datetime"}
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        return tags

feature_engine/discretisation/arbitrary.py

@@ -208,3 +208,7 @@ def _more_tags(self):
             "check_parameters_default_constructible"
         ] = "transformer has 1 mandatory parameter"
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        return tags

feature_engine/discretisation/decision_tree.py

@@ -345,3 +345,7 @@ def _more_tags(self):
         tags_dict["variables"] = "numerical"
         tags_dict["requires_y"] = True
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        return tags

feature_engine/encoding/base_encoder.py

@@ -102,7 +102,7 @@ def __init__(
         self.missing_values = missing_values
 
 
-class CategoricalMethodsMixin(BaseEstimator, TransformerMixin, GetFeatureNamesOutMixin):
+class CategoricalMethodsMixin(TransformerMixin, BaseEstimator, GetFeatureNamesOutMixin):
     """Shared methods across categorical transformers.
 
     - BaseEstimator brings methods get_params() and set_params().
@@ -299,3 +299,7 @@ def _more_tags(self):
         # so we need to leave without this test
         tags_dict["_xfail_checks"]["check_estimators_nan_inf"] = "transformer allows NA"
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        return tags

feature_engine/encoding/count_frequency.py

@@ -49,7 +49,7 @@
     transform=_transform_encoders_docstring,
     inverse_transform=_inverse_transform_docstring,
 )
-class CountFrequencyEncoder(CategoricalInitMixinNA, CategoricalMethodsMixin):
+class CountFrequencyEncoder(CategoricalMethodsMixin, CategoricalInitMixinNA):
     """
     The CountFrequencyEncoder() replaces categories by either the count or the
     percentage of observations per category.

feature_engine/encoding/decision_tree.py

@@ -52,7 +52,7 @@
     fit_transform=_fit_transform_docstring,
     inverse_transform=_inverse_transform_docstring,
 )
-class DecisionTreeEncoder(CategoricalInitMixin, CategoricalMethodsMixin):
+class DecisionTreeEncoder(CategoricalMethodsMixin, CategoricalInitMixin):
     """
     The DecisionTreeEncoder() encodes categorical variables with the predictions
     of a decision tree.

feature_engine/encoding/mean_encoding.py

@@ -48,7 +48,7 @@
     transform=_transform_encoders_docstring,
     inverse_transform=_inverse_transform_docstring,
 )
-class MeanEncoder(CategoricalInitMixinNA, CategoricalMethodsMixin):
+class MeanEncoder(CategoricalMethodsMixin, CategoricalInitMixinNA):
     """
     The MeanEncoder() replaces categories by the mean value of the target for each
     category.
@@ -266,3 +266,7 @@ def _more_tags(self):
         tags_dict = super()._more_tags()
         tags_dict["requires_y"] = True
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        return tags

feature_engine/encoding/one_hot.py

@@ -32,7 +32,7 @@
     n_features_in_=_n_features_in_docstring,
     fit_transform=_fit_transform_docstring,
 )
-class OneHotEncoder(CategoricalInitMixin, CategoricalMethodsMixin):
+class OneHotEncoder(CategoricalMethodsMixin, CategoricalInitMixin):
     """
     The OneHotEncoder() replaces categorical variables by a set of binary variables
     representing each one of the unique categories in the variable.

feature_engine/encoding/ordinal.py

@@ -48,7 +48,7 @@
     transform=_transform_encoders_docstring,
     inverse_transform=_inverse_transform_docstring,
 )
-class OrdinalEncoder(CategoricalInitMixinNA, CategoricalMethodsMixin):
+class OrdinalEncoder(CategoricalMethodsMixin, CategoricalInitMixinNA):
     """
     The OrdinalEncoder() replaces categories by ordinal numbers
     (0, 1, 2, 3, etc). The numbers can be ordered based on the mean of the target

feature_engine/encoding/rare_label.py

@@ -35,7 +35,7 @@
     n_features_in_=_n_features_in_docstring,
     fit_transform=_fit_transform_docstring,
 )
-class RareLabelEncoder(CategoricalInitMixinNA, CategoricalMethodsMixin):
+class RareLabelEncoder(CategoricalMethodsMixin, CategoricalInitMixinNA):
     """
     The RareLabelEncoder() groups rare or infrequent categories in
     a new category called "Rare", or any other name entered by the user.

feature_engine/encoding/similarity_encoder.py

@@ -38,7 +38,7 @@ def _gpm_fast(x1: str, x2: str) -> float:
     n_features_in_=_n_features_in_docstring,
     fit_transform=_fit_transform_docstring,
 )
-class StringSimilarityEncoder(CategoricalInitMixin, CategoricalMethodsMixin):
+class StringSimilarityEncoder(CategoricalMethodsMixin, CategoricalInitMixin):
     """
     The StringSimilarityEncoder() replaces categorical variables with a set of float
     variables that capture the similarity between the category names. The new variables

feature_engine/encoding/woe.py

@@ -94,7 +94,7 @@ def _calculate_woe(
     transform=_transform_encoders_docstring,
     inverse_transform=_inverse_transform_docstring,
 )
-class WoEEncoder(CategoricalInitMixin, CategoricalMethodsMixin, WoE):
+class WoEEncoder(CategoricalMethodsMixin, CategoricalInitMixin, WoE):
     """
     The WoEEncoder() replaces categories by the weight of evidence
     (WoE). The WoE was used primarily in the financial sector to create credit risk
@@ -284,8 +284,12 @@ def _more_tags(self):
         # in the current format, the tests are performed using continuous np.arrays
         # this means that when we encode some of the values, the denominator is 0
         # and this the transformer raises an error, and the test fails.
-        # For this reason, most sklearn transformers will fail. And it has nothing to
+        # For this reason, most sklearn tests will fail. And it has nothing to
         # do with the class not being compatible, it is just that the inputs passed
         # are not suitable
         tags_dict["_skip_test"] = True
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        return tags

feature_engine/imputation/base_imputer.py

@@ -7,7 +7,7 @@
 from feature_engine.tags import _return_tags
 
 
-class BaseImputer(BaseEstimator, TransformerMixin, GetFeatureNamesOutMixin):
+class BaseImputer(TransformerMixin, BaseEstimator, GetFeatureNamesOutMixin):
     """shared set-up checks and methods across imputers"""
 
     def _transform(self, X: pd.DataFrame) -> pd.DataFrame:
@@ -78,3 +78,8 @@ def _more_tags(self):
         tags_dict["allow_nan"] = True
         tags_dict["variables"] = "numerical"
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        tags.input_tags.allow_nan = True
+        return tags

feature_engine/imputation/categorical.py

@@ -257,3 +257,8 @@ def _more_tags(self):
         tags_dict["allow_nan"] = True
         tags_dict["variables"] = "categorical"
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        tags.input_tags.allow_nan = True
+        return tags

feature_engine/imputation/drop_missing_data.py

@@ -223,3 +223,8 @@ def _more_tags(self):
         tags_dict["allow_nan"] = True
         tags_dict["variables"] = "all"
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        tags.input_tags.allow_nan = True
+        return tags

feature_engine/imputation/missing_indicator.py

@@ -179,3 +179,8 @@ def _more_tags(self):
         tags_dict["allow_nan"] = True
         tags_dict["variables"] = "all"
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        tags.input_tags.allow_nan = True
+        return tags

feature_engine/imputation/random_sample.py

@@ -275,3 +275,8 @@ def _more_tags(self):
         tags_dict["allow_nan"] = True
         tags_dict["variables"] = "all"
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        tags.input_tags.allow_nan = True
+        return tags

feature_engine/outliers/artbitrary.py

@@ -204,3 +204,7 @@ def _more_tags(self):
             "check_parameters_default_constructible"
         ] = "transformer has 1 mandatory parameter"
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        return tags

feature_engine/outliers/base_outlier.py

@@ -21,7 +21,7 @@
 )
 
 
-class BaseOutlier(BaseEstimator, TransformerMixin, GetFeatureNamesOutMixin):
+class BaseOutlier(TransformerMixin, BaseEstimator, GetFeatureNamesOutMixin):
     """shared set-up checks and methods across outlier transformers"""
 
     def _check_transform_input_and_state(self, X: pd.DataFrame) -> pd.DataFrame:
@@ -96,6 +96,10 @@ def _more_tags(self):
         tags_dict["variables"] = "numerical"
         return tags_dict
 
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        return tags
+
 
 class WinsorizerBase(BaseOutlier):
 
@@ -289,7 +293,7 @@ def _more_tags(self):
         tags_dict = _return_tags()
         tags_dict["variables"] = "numerical"
         # =======  this tests fail because the transformers throw an error
-        # when variance of the any input feature is 0.
+        # when variance of any input feature is 0.
         # Nothing to do with the test itself but
         # mostly with the data created and used in the test
         msg = (
@@ -298,3 +302,7 @@ def _more_tags(self):
         )
         tags_dict["_xfail_checks"]["check_fit2d_1sample"] = msg
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        return tags

feature_engine/preprocessing/match_categories.py

@@ -31,7 +31,7 @@
     n_features_in_=_n_features_in_docstring,
 )
 class MatchCategories(
-    CategoricalInitMixinNA, CategoricalMethodsMixin, GetFeatureNamesOutMixin
+    CategoricalMethodsMixin, CategoricalInitMixinNA, GetFeatureNamesOutMixin
 ):
     """
     MatchCategories() ensures that categorical variables are encoded as pandas

feature_engine/preprocessing/match_columns.py

@@ -10,7 +10,7 @@
 from feature_engine.tags import _return_tags
 
 
-class MatchVariables(BaseEstimator, TransformerMixin, GetFeatureNamesOutMixin):
+class MatchVariables(TransformerMixin, BaseEstimator, GetFeatureNamesOutMixin):
     """
     MatchVariables() ensures that the same variables observed in the train set
     are present in the test set. If the dataset to transform contains variables that
@@ -297,3 +297,7 @@ def _more_tags(self):
         tags_dict["_xfail_checks"]["check_estimators_nan_inf"] = msg
 
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        return tags

feature_engine/selection/base_recursive_selector.py

@@ -209,3 +209,7 @@ def _more_tags(self):
         tags_dict["_xfail_checks"]["check_fit2d_1feature"] = msg
 
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        return tags

feature_engine/selection/base_selector.py

@@ -8,7 +8,7 @@
 from feature_engine.tags import _return_tags
 
 
-class BaseSelector(BaseEstimator, TransformerMixin, GetFeatureNamesOutMixin):
+class BaseSelector(TransformerMixin, BaseEstimator, GetFeatureNamesOutMixin):
     """
     Shared set-up checks and methods across selectors.
 
@@ -124,3 +124,7 @@ def _more_tags(self):
         tags_dict["_xfail_checks"]["check_fit2d_1feature"] = msg
 
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        return tags

feature_engine/selection/drop_constant_features.py

@@ -228,3 +228,8 @@ def _more_tags(self):
             "check_fit2d_1sample"
         ] = "the transformer raises an error when dropping all columns, ok to fail"
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        tags.input_tags.allow_nan = True
+        return tags

feature_engine/selection/drop_duplicate_features.py

@@ -187,3 +187,8 @@ def _more_tags(self):
         tags_dict["_xfail_checks"]["check_fit2d_1feature"] = msg
 
         return tags_dict
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        tags.input_tags.allow_nan = True
+        return tags