FIX forward properly the metadata with the pipeline (#1115)

* FIX forward properly the metadata with the pipeline

* older sklearn

Author: glemaitre

imblearn/base.py

@@ -22,6 +22,14 @@
     METHODS.append("fit_transform")
 METHODS.append("fit_resample")
 
+try:
+    from sklearn.utils._metadata_requests import SIMPLE_METHODS
+
+    SIMPLE_METHODS.append("fit_resample")
+except ImportError:
+    # in older versions of scikit-learn, only METHODS is used
+    pass
+
 
 class SamplerMixin(metaclass=ABCMeta):
     """Mixin class for samplers with abstract method.
@@ -33,7 +41,7 @@ class SamplerMixin(metaclass=ABCMeta):
     _estimator_type = "sampler"
 
     @_fit_context(prefer_skip_nested_validation=True)
-    def fit(self, X, y):
+    def fit(self, X, y, **params):
         """Check inputs and statistics of the sampler.
 
         You should use ``fit_resample`` in all cases.
@@ -47,6 +55,9 @@ def fit(self, X, y):
         y : array-like of shape (n_samples,)
             Target array.
 
+        **params : dict
+            Extra parameters to use by the sampler.
+
         Returns
         -------
         self : object
@@ -58,7 +69,8 @@ def fit(self, X, y):
         )
         return self
 
-    def fit_resample(self, X, y):
+    @_fit_context(prefer_skip_nested_validation=True)
+    def fit_resample(self, X, y, **params):
         """Resample the dataset.
 
         Parameters
@@ -70,6 +82,9 @@ def fit_resample(self, X, y):
         y : array-like of shape (n_samples,)
             Corresponding label for each sample in X.
 
+        **params : dict
+            Extra parameters to use by the sampler.
+
         Returns
         -------
         X_resampled : {array-like, dataframe, sparse matrix} of shape \
@@ -87,7 +102,7 @@ def fit_resample(self, X, y):
             self.sampling_strategy, y, self._sampling_type
         )
 
-        output = self._fit_resample(X, y)
+        output = self._fit_resample(X, y, **params)
 
         y_ = (
             label_binarize(output[1], classes=np.unique(y)) if binarize_y else output[1]
@@ -97,7 +112,7 @@ def fit_resample(self, X, y):
         return (X_, y_) if len(output) == 2 else (X_, y_, output[2])
 
     @abstractmethod
-    def _fit_resample(self, X, y):
+    def _fit_resample(self, X, y, **params):
         """Base method defined in each sampler to defined the sampling
         strategy.
 
@@ -109,6 +124,9 @@ def _fit_resample(self, X, y):
         y : array-like of shape (n_samples,)
             Corresponding label for each sample in X.
 
+        **params : dict
+            Extra parameters to use by the sampler.
+
         Returns
         -------
         X_resampled : {ndarray, sparse matrix} of shape \
@@ -139,7 +157,7 @@ def _check_X_y(self, X, y, accept_sparse=None):
         X, y = validate_data(self, X=X, y=y, reset=True, accept_sparse=accept_sparse)
         return X, y, binarize_y
 
-    def fit(self, X, y):
+    def fit(self, X, y, **params):
         """Check inputs and statistics of the sampler.
 
         You should use ``fit_resample`` in all cases.
@@ -158,10 +176,9 @@ def fit(self, X, y):
         self : object
             Return the instance itself.
         """
-        self._validate_params()
-        return super().fit(X, y)
+        return super().fit(X, y, **params)
 
-    def fit_resample(self, X, y):
+    def fit_resample(self, X, y, **params):
         """Resample the dataset.
 
         Parameters
@@ -182,8 +199,7 @@ def fit_resample(self, X, y):
         y_resampled : array-like of shape (n_samples_new,)
             The corresponding label of `X_resampled`.
         """
-        self._validate_params()
-        return super().fit_resample(X, y)
+        return super().fit_resample(X, y, **params)
 
     def _more_tags(self):
         return {"X_types": ["2darray", "sparse", "dataframe"]}

imblearn/pipeline.py

@@ -1168,34 +1168,45 @@ def get_metadata_routing(self):
         router = MetadataRouter(owner=self.__class__.__name__)
 
         # first we add all steps except the last one
-        for _, name, trans in self._iter(with_final=False, filter_passthrough=True):
+        for _, name, trans in self._iter(
+            with_final=False, filter_passthrough=True, filter_resample=False
+        ):
             method_mapping = MethodMapping()
             # fit, fit_predict, and fit_transform call fit_transform if it
             # exists, or else fit and transform
             if hasattr(trans, "fit_transform"):
-                method_mapping.add(caller="fit", callee="fit_transform")
-                method_mapping.add(caller="fit_transform", callee="fit_transform")
-                method_mapping.add(caller="fit_predict", callee="fit_transform")
-                method_mapping.add(caller="fit_resample", callee="fit_transform")
+                (
+                    method_mapping.add(caller="fit", callee="fit_transform")
+                    .add(caller="fit_transform", callee="fit_transform")
+                    .add(caller="fit_predict", callee="fit_transform")
+                )
             else:
-                method_mapping.add(caller="fit", callee="fit")
-                method_mapping.add(caller="fit", callee="transform")
-                method_mapping.add(caller="fit_transform", callee="fit")
-                method_mapping.add(caller="fit_transform", callee="transform")
-                method_mapping.add(caller="fit_predict", callee="fit")
-                method_mapping.add(caller="fit_predict", callee="transform")
-                method_mapping.add(caller="fit_resample", callee="fit")
-                method_mapping.add(caller="fit_resample", callee="transform")
-
-            method_mapping.add(caller="predict", callee="transform")
-            method_mapping.add(caller="predict", callee="transform")
-            method_mapping.add(caller="predict_proba", callee="transform")
-            method_mapping.add(caller="decision_function", callee="transform")
-            method_mapping.add(caller="predict_log_proba", callee="transform")
-            method_mapping.add(caller="transform", callee="transform")
-            method_mapping.add(caller="inverse_transform", callee="inverse_transform")
-            method_mapping.add(caller="score", callee="transform")
-            method_mapping.add(caller="fit_resample", callee="transform")
+                (
+                    method_mapping.add(caller="fit", callee="fit")
+                    .add(caller="fit", callee="transform")
+                    .add(caller="fit_transform", callee="fit")
+                    .add(caller="fit_transform", callee="transform")
+                    .add(caller="fit_predict", callee="fit")
+                    .add(caller="fit_predict", callee="transform")
+                )
+
+            (
+                # handling sampler if the fit_* stage
+                method_mapping.add(caller="fit", callee="fit_resample")
+                .add(caller="fit_transform", callee="fit_resample")
+                .add(caller="fit_predict", callee="fit_resample")
+            )
+            (
+                method_mapping.add(caller="predict", callee="transform")
+                .add(caller="predict", callee="transform")
+                .add(caller="predict_proba", callee="transform")
+                .add(caller="decision_function", callee="transform")
+                .add(caller="predict_log_proba", callee="transform")
+                .add(caller="transform", callee="transform")
+                .add(caller="inverse_transform", callee="inverse_transform")
+                .add(caller="score", callee="transform")
+                .add(caller="fit_resample", callee="transform")
+            )
 
             router.add(method_mapping=method_mapping, **{name: trans})
 
@@ -1207,23 +1218,24 @@ def get_metadata_routing(self):
         method_mapping = MethodMapping()
         if hasattr(final_est, "fit_transform"):
             method_mapping.add(caller="fit_transform", callee="fit_transform")
-            method_mapping.add(caller="fit_resample", callee="fit_transform")
         else:
+            (
+                method_mapping.add(caller="fit", callee="fit").add(
+                    caller="fit", callee="transform"
+                )
+            )
+        (
             method_mapping.add(caller="fit", callee="fit")
-            method_mapping.add(caller="fit", callee="transform")
-            method_mapping.add(caller="fit_resample", callee="fit")
-            method_mapping.add(caller="fit_resample", callee="transform")
-
-        method_mapping.add(caller="fit", callee="fit")
-        method_mapping.add(caller="predict", callee="predict")
-        method_mapping.add(caller="fit_predict", callee="fit_predict")
-        method_mapping.add(caller="predict_proba", callee="predict_proba")
-        method_mapping.add(caller="decision_function", callee="decision_function")
-        method_mapping.add(caller="predict_log_proba", callee="predict_log_proba")
-        method_mapping.add(caller="transform", callee="transform")
-        method_mapping.add(caller="inverse_transform", callee="inverse_transform")
-        method_mapping.add(caller="score", callee="score")
-        method_mapping.add(caller="fit_resample", callee="fit_resample")
+            .add(caller="predict", callee="predict")
+            .add(caller="fit_predict", callee="fit_predict")
+            .add(caller="predict_proba", callee="predict_proba")
+            .add(caller="decision_function", callee="decision_function")
+            .add(caller="predict_log_proba", callee="predict_log_proba")
+            .add(caller="transform", callee="transform")
+            .add(caller="inverse_transform", callee="inverse_transform")
+            .add(caller="score", callee="score")
+            .add(caller="fit_resample", callee="fit_resample")
+        )
 
         router.add(method_mapping=method_mapping, **{final_name: final_est})
         return router

imblearn/tests/test_pipeline.py

@@ -34,6 +34,7 @@
 )
 from sklearn.utils.fixes import parse_version
 
+from imblearn.base import BaseSampler
 from imblearn.datasets import make_imbalance
 from imblearn.pipeline import Pipeline, make_pipeline
 from imblearn.under_sampling import EditedNearestNeighbours as ENN
@@ -1495,3 +1496,24 @@ def test_transform_input_sklearn_version():
 
 # end of transform_input tests
 # =============================
+
+
+def test_metadata_routing_with_sampler():
+    """Check that we can use a sampler with metadata routing."""
+    X, y = make_classification()
+    cost_matrix = np.random.rand(X.shape[0], 2, 2)
+
+    class CostSensitiveSampler(BaseSampler):
+        def fit_resample(self, X, y, cost_matrix=None):
+            return self._fit_resample(X, y, cost_matrix=cost_matrix)
+
+        def _fit_resample(self, X, y, cost_matrix=None):
+            self.cost_matrix_ = cost_matrix
+            return X, y
+
+    with config_context(enable_metadata_routing=True):
+        sampler = CostSensitiveSampler().set_fit_resample_request(cost_matrix=True)
+        pipeline = Pipeline([("sampler", sampler), ("model", LogisticRegression())])
+        pipeline.fit(X, y, cost_matrix=cost_matrix)
+
+        assert_allclose(pipeline[0].cost_matrix_, cost_matrix)