scikit-learn · glemaitre · Oct 30, 2023 · Sep 28, 2023 · Sep 28, 2023 · Sep 29, 2023
diff --git a/doc/metadata_routing.rst b/doc/metadata_routing.rst
@@ -252,6 +252,7 @@ Meta-estimators and functions supporting metadata routing:
 
 - :class:`sklearn.calibration.CalibratedClassifierCV`
 - :class:`sklearn.compose.ColumnTransformer`
+- :class:`sklearn.feature_selection.SelectFromModel`
 - :class:`sklearn.linear_model.ElasticNetCV`
 - :class:`sklearn.linear_model.LarsCV`
 - :class:`sklearn.linear_model.LassoCV`
@@ -290,7 +291,6 @@ Meta-estimators and tools not supporting metadata routing yet:
 - :class:`sklearn.ensemble.VotingRegressor`
 - :class:`sklearn.feature_selection.RFE`
 - :class:`sklearn.feature_selection.RFECV`
-- :class:`sklearn.feature_selection.SelectFromModel`
 - :class:`sklearn.feature_selection.SequentialFeatureSelector`
 - :class:`sklearn.impute.IterativeImputer`
 - :class:`sklearn.linear_model.RANSACRegressor`

diff --git a/doc/whats_new/v1.4.rst b/doc/whats_new/v1.4.rst
@@ -107,6 +107,10 @@ more details.
   ``**score_params`` which are passed to the underlying scorer.
   :pr:`26525` by :user:`Omar Salman <OmarManzoor>`.
 
+- |Feature| :class:`feature_selection.SelectFromModel` now supports metadata
+  routing in `fit` and `partial_fit`.
+  :pr:`27490` by :user:`Stefanie Senger <StefanieSenger>`.
+
 - |Feature| :class:`linear_model.OrthogonalMatchingPursuitCV` now supports
   metadata routing. Its `fit` now accepts ``**fit_params``, which are passed to
   the underlying splitter. :pr:`27500` by :user:`Stefanie Senger

diff --git a/sklearn/feature_selection/_from_model.py b/sklearn/feature_selection/_from_model.py
@@ -11,8 +11,10 @@
 from ..utils._param_validation import HasMethods, Interval, Options
 from ..utils._tags import _safe_tags
 from ..utils.metadata_routing import (
-    _raise_for_unsupported_routing,
-    _RoutingNotSupportedMixin,
+    MetadataRouter,
+    MethodMapping,
+    _routing_enabled,
+    process_routing,
 )
 from ..utils.metaestimators import available_if
 from ..utils.validation import _num_features, check_is_fitted, check_scalar
@@ -82,9 +84,7 @@ def _estimator_has(attr):
     )
 
 
-class SelectFromModel(
-    _RoutingNotSupportedMixin, MetaEstimatorMixin, SelectorMixin, BaseEstimator
-):
+class SelectFromModel(MetaEstimatorMixin, SelectorMixin, BaseEstimator):
     """Meta-transformer for selecting features based on importance weights.
 
     .. versionadded:: 0.17
@@ -341,14 +341,25 @@ def fit(self, X, y=None, **fit_params):
             classification, real numbers in regression).
 
         **fit_params : dict
-            Other estimator specific parameters.
+            - If `enable_metadata_routing=False` (default):
+
+                Parameters directly passed to the `partial_fit` method of the
+                sub-estimator. They are ignored if `prefit=True`.
+
+            - If `enable_metadata_routing=True`:
+
+                Parameters safely routed to the `partial_fit` method of the
+                sub-estimator. They are ignored if `prefit=True`.
+
+                .. versionchanged:: 1.4
+                    See :ref:`Metadata Routing User Guide <metadata_routing>` for
+                    more details.
 
         Returns
         -------
         self : object
             Fitted estimator.
         """
-        _raise_for_unsupported_routing(self, "fit", **fit_params)
         self._check_max_features(X)
 
         if self.prefit:
@@ -361,8 +372,14 @@ def fit(self, X, y=None, **fit_params):
                 ) from exc
             self.estimator_ = deepcopy(self.estimator)
         else:
-            self.estimator_ = clone(self.estimator)
-            self.estimator_.fit(X, y, **fit_params)
+            if _routing_enabled():
+                routed_params = process_routing(self, "fit", **fit_params)
+                self.estimator_ = clone(self.estimator)
+                self.estimator_.fit(X, y, **routed_params.estimator.fit)
+            else:
+                # TODO(SLEP6): remove when metadata routing cannot be disabled.
+                self.estimator_ = clone(self.estimator)
+                self.estimator_.fit(X, y, **fit_params)
 
         if hasattr(self.estimator_, "feature_names_in_"):
             self.feature_names_in_ = self.estimator_.feature_names_in_
@@ -387,7 +404,7 @@ def threshold_(self):
         # SelectFromModel.estimator is not validated yet
         prefer_skip_nested_validation=False
     )
-    def partial_fit(self, X, y=None, **fit_params):
+    def partial_fit(self, X, y=None, **partial_fit_params):
         """Fit the SelectFromModel meta-transformer only once.
 
         Parameters
@@ -399,8 +416,24 @@ def partial_fit(self, X, y=None, **fit_params):
             The target values (integers that correspond to classes in
             classification, real numbers in regression).
 
-        **fit_params : dict
-            Other estimator specific parameters.
+        **partial_fit_params : dict
+            - If `enable_metadata_routing=False` (default):
+
+                Parameters directly passed to the `partial_fit` method of the
+                sub-estimator.
+
+            - If `enable_metadata_routing=True`:
+
+                Parameters passed to the `partial_fit` method of the
+                sub-estimator. They are ignored if `prefit=True`.
+
+                .. versionchanged:: 1.4
+                    `**partial_fit_params` are routed to the sub-estimator, if
+                    `enable_metadata_routing=True` is set via
+                    :func:`~sklearn.set_config`, which allows for aliasing.
-                    `**partial_fit_params` are routed to the sub-estimator, if
-                    `enable_metadata_routing=True` is set via
-                    :func:`~sklearn.set_config`, which allows for aliasing.
+                    Only available if `enable_metadata_routing=True`,
+                    which can be set by using
+                    ``sklearn.set_config(enable_metadata_routing=True)``.
+                    See :ref:`Metadata Routing User Guide <metadata_routing>` for
+                    more details.
-                    `**partial_fit_params` are routed to the sub-estimator, if
-                    `enable_metadata_routing=True` is set via
-                    :func:`~sklearn.set_config`, which allows for aliasing.
+                    Only available if `enable_metadata_routing=True`,
+                    which can be set by using
+                    ``sklearn.set_config(enable_metadata_routing=True)``.
+                    See :ref:`Metadata Routing User Guide <metadata_routing>` for
+                    more details.
+
+                See :ref:`Metadata Routing User Guide <metadata_routing>` for
+                more details.
 
         Returns
         -------
@@ -426,7 +459,13 @@ def partial_fit(self, X, y=None, **fit_params):
 
         if first_call:
             self.estimator_ = clone(self.estimator)
-        self.estimator_.partial_fit(X, y, **fit_params)
+        if _routing_enabled():
+            routed_params = process_routing(self, "partial_fit", **partial_fit_params)
+            self.estimator_ = clone(self.estimator)
+            self.estimator_.partial_fit(X, y, **routed_params.estimator.partial_fit)
+        else:
+            # TODO(SLEP6): remove when metadata routing cannot be disabled.
+            self.estimator_.partial_fit(X, y, **partial_fit_params)
 
         if hasattr(self.estimator_, "feature_names_in_"):
             self.feature_names_in_ = self.estimator_.feature_names_in_
@@ -451,5 +490,27 @@ def n_features_in_(self):
 
         return self.estimator_.n_features_in_
 
+    def get_metadata_routing(self):
+        """Get metadata routing of this object.
+
+        Please check :ref:`User Guide <metadata_routing>` on how the routing
+        mechanism works.
+
+        .. versionadded:: 1.4
+
+        Returns
+        -------
+        routing : MetadataRouter
+            A :class:`~sklearn.utils.metadata_routing.MetadataRouter` encapsulating
+            routing information.
+        """
+        router = MetadataRouter(owner=self.__class__.__name__).add(
+            estimator=self.estimator,
+            method_mapping=MethodMapping()
+            .add(callee="partial_fit", caller="partial_fit")
+            .add(callee="fit", caller="fit"),
+        )
+        return router
+
     def _more_tags(self):
         return {"allow_nan": _safe_tags(self.estimator, key="allow_nan")}
diff --git a/sklearn/tests/test_metaestimators_metadata_routing.py b/sklearn/tests/test_metaestimators_metadata_routing.py
@@ -5,6 +5,7 @@
 import pytest
 
 from sklearn import config_context
+from sklearn.base import is_classifier
 from sklearn.calibration import CalibratedClassifierCV
 from sklearn.compose import TransformedTargetRegressor
 from sklearn.covariance import GraphicalLassoCV
@@ -228,6 +229,15 @@ def enable_slep006():
         "y": y,
         "estimator_routing_methods": ["fit"],
     },
+    {
+        "metaestimator": SelectFromModel,
+        "estimator_name": "estimator",
+        "estimator": ConsumingClassifier,
+        "X": X,
+        "y": y,
+        "estimator_routing_methods": ["fit", "partial_fit"],
+        "method_args": {"partial_fit": {"classes": classes}},
+    },
     {
         "metaestimator": OrthogonalMatchingPursuitCV,
         "X": X,
@@ -325,7 +335,6 @@ def enable_slep006():
     RFECV(ConsumingClassifier()),
     RidgeCV(),
     RidgeClassifierCV(),
-    SelectFromModel(ConsumingClassifier()),
     SelfTrainingClassifier(ConsumingClassifier()),
     SequentialFeatureSelector(ConsumingClassifier()),
     StackingClassifier(ConsumingClassifier()),
@@ -477,6 +486,8 @@ def set_request(estimator, method_name):
         # e.g. call set_fit_request on estimator
         set_request_for_method = getattr(estimator, f"set_{method_name}_request")
         set_request_for_method(sample_weight=True, metadata=True)
+        if is_classifier(estimator) and method_name == "partial_fit":
+            set_request_for_method(classes=True)
 
     cls = metaestimator["metaestimator"]
     X = metaestimator["X"]