iter

Author: glemaitre

.github/workflows/tests.yml

@@ -18,8 +18,8 @@ jobs:
             ci-py310-min-optional-dependencies,
             ci-py310-min-keras,
             ci-py310-min-tensorflow,
-            ci-py311-sklearn-1-3,
             ci-py311-sklearn-1-4,
+            ci-py311-sklearn-1-5,
             ci-py311-latest-keras,
             ci-py311-latest-tensorflow,
             ci-py313-latest-dependencies,

imblearn/ensemble/_easy_ensemble.py

@@ -315,7 +315,9 @@ def decision_function(self, X):
             X=X,
             accept_sparse=["csr", "csc"],
             dtype=None,
-            ensure_all_finite=False,
+            ensure_all_finite=(
+                "allow_nan" if get_tags(self).input_tags.allow_nan else True
+            ),
             reset=False,
         )
 
@@ -352,7 +354,6 @@ def _get_estimator(self):
                 return AdaBoostClassifier()
         return self.estimator
 
-    # TODO: remove when minimum supported version of scikit-learn is 1.5
     def _more_tags(self):
         return {"allow_nan": _safe_tags(self._get_estimator(), "allow_nan")}
 

imblearn/ensemble/tests/test_easy_ensemble.py

@@ -6,7 +6,7 @@
 import numpy as np
 import pytest
 from sklearn.datasets import load_iris, make_hastie_10_2
-from sklearn.ensemble import AdaBoostClassifier
+from sklearn.ensemble import AdaBoostClassifier, GradientBoostingClassifier
 from sklearn.feature_selection import SelectKBest
 from sklearn.model_selection import GridSearchCV, train_test_split
 from sklearn.utils._testing import assert_allclose, assert_array_equal
@@ -41,8 +41,8 @@
 @pytest.mark.parametrize(
     "estimator",
     [
-        AdaBoostClassifier(n_estimators=5),
-        AdaBoostClassifier(n_estimators=10),
+        GradientBoostingClassifier(n_estimators=5),
+        GradientBoostingClassifier(n_estimators=10),
     ],
 )
 def test_easy_ensemble_classifier(n_estimators, estimator):
@@ -89,10 +89,10 @@ def test_estimator():
     assert isinstance(ensemble.estimator_.steps[-1][1], AdaBoostClassifier)
 
     ensemble = EasyEnsembleClassifier(
-        2, AdaBoostClassifier(), n_jobs=-1, random_state=0
+        2, GradientBoostingClassifier(), n_jobs=-1, random_state=0
     ).fit(X_train, y_train)
 
-    assert isinstance(ensemble.estimator_.steps[-1][1], AdaBoostClassifier)
+    assert isinstance(ensemble.estimator_.steps[-1][1], GradientBoostingClassifier)
 
 
 def test_bagging_with_pipeline():
@@ -104,7 +104,7 @@ def test_bagging_with_pipeline():
     )
     estimator = EasyEnsembleClassifier(
         n_estimators=2,
-        estimator=make_pipeline(SelectKBest(k=1), AdaBoostClassifier()),
+        estimator=make_pipeline(SelectKBest(k=1), GradientBoostingClassifier()),
     )
     estimator.fit(X, y).predict(X)
 
@@ -196,7 +196,7 @@ def test_easy_ensemble_classifier_single_estimator():
     clf1 = EasyEnsembleClassifier(n_estimators=1, random_state=0).fit(X_train, y_train)
     clf2 = make_pipeline(
         RandomUnderSampler(random_state=0),
-        AdaBoostClassifier(random_state=0),
+        GradientBoostingClassifier(random_state=0),
     ).fit(X_train, y_train)
 
     assert_array_equal(clf1.predict(X_test), clf2.predict(X_test))
@@ -215,7 +215,7 @@ def test_easy_ensemble_classifier_grid_search():
         "estimator__n_estimators": [3, 4],
     }
     grid_search = GridSearchCV(
-        EasyEnsembleClassifier(estimator=AdaBoostClassifier()),
+        EasyEnsembleClassifier(estimator=GradientBoostingClassifier()),
         parameters,
         cv=5,
     )

imblearn/pipeline.py

@@ -15,6 +15,7 @@
 # License: BSD
 import warnings
 from contextlib import contextmanager
+from copy import deepcopy
 
 import sklearn
 from sklearn import pipeline
@@ -25,10 +26,8 @@
     METHODS,
     MetadataRouter,
     MethodMapping,
-    _raise_for_params,
     _routing_enabled,
     get_routing_for_object,
-    process_routing,
 )
 from sklearn.utils._param_validation import HasMethods
 from sklearn.utils.fixes import parse_version
@@ -38,9 +37,14 @@
 from .utils._sklearn_compat import (
     _fit_context,
     _print_elapsed_time,
+    _raise_for_params,
+    get_tags,
+    process_routing,
     validate_params,
 )
 
+if "fit_predict" not in METHODS:
+    METHODS.append("fit_predict")
 METHODS.append("fit_resample")
 
 __all__ = ["Pipeline", "make_pipeline"]
@@ -245,6 +249,12 @@ class Pipeline(pipeline.Pipeline):
         "verbose": ["boolean"],
     }
 
+    def __init__(self, steps, *, transform_input=None, memory=None, verbose=False):
+        self.steps = steps
+        self.transform_input = transform_input
+        self.memory = memory
+        self.verbose = verbose
+
     # BaseEstimator interface
 
     def _validate_steps(self):
@@ -1162,35 +1172,29 @@ def get_metadata_routing(self):
             # 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")
-                    .add(caller="fit_transform", callee="fit_transform")
-                    .add(caller="fit_predict", callee="fit_transform")
-                    .add(caller="fit_resample", callee="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")
             else:
-                (
-                    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")
-                    .add(caller="fit_resample", callee="fit")
-                    .add(caller="fit_resample", callee="transform")
-                )
-
-            (
-                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")
-            )
+                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")
 
             router.add(method_mapping=method_mapping, **{name: trans})
 
@@ -1201,30 +1205,24 @@ def get_metadata_routing(self):
         # then we add the last step
         method_mapping = MethodMapping()
         if hasattr(final_est, "fit_transform"):
-            (
-                method_mapping.add(caller="fit_transform", callee="fit_transform").add(
-                    caller="fit_resample", callee="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")
-                .add(caller="fit_resample", callee="fit")
-                .add(caller="fit_resample", callee="transform")
-            )
-        (
             method_mapping.add(caller="fit", callee="fit")
-            .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")
-        )
+            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")
 
         router.add(method_mapping=method_mapping, **{final_name: final_est})
         return router
@@ -1258,6 +1256,67 @@ def _check_method_params(self, method, props, **kwargs):
                 fit_params_steps[step]["fit_predict"][param] = pval
             return fit_params_steps
 
+    def __sklearn_is_fitted__(self):
+        """Indicate whether pipeline has been fit.
+
+        This is done by checking whether the last non-`passthrough` step of the
+        pipeline is fitted.
+
+        An empty pipeline is considered fitted.
+        """
+
+        # First find the last step that is not 'passthrough'
+        last_step = None
+        for _, estimator in reversed(self.steps):
+            if estimator != "passthrough":
+                last_step = estimator
+                break
+
+        if last_step is None:
+            # All steps are 'passthrough', so the pipeline is considered fitted
+            return True
+
+        try:
+            # check if the last step of the pipeline is fitted
+            # we only check the last step since if the last step is fit, it
+            # means the previous steps should also be fit. This is faster than
+            # checking if every step of the pipeline is fit.
+            check_is_fitted(last_step)
+            return True
+        except NotFittedError:
+            return False
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+
+        if not self.steps:
+            return tags
+
+        try:
+            if self.steps[0][1] is not None and self.steps[0][1] != "passthrough":
+                tags.input_tags.pairwise = get_tags(
+                    self.steps[0][1]
+                ).input_tags.pairwise
+        except (ValueError, AttributeError, TypeError):
+            # This happens when the `steps` is not a list of (name, estimator)
+            # tuples and `fit` is not called yet to validate the steps.
+            pass
+
+        try:
+            if self.steps[-1][1] is not None and self.steps[-1][1] != "passthrough":
+                last_step_tags = get_tags(self.steps[-1][1])
+                tags.estimator_type = last_step_tags.estimator_type
+                tags.target_tags.multi_output = last_step_tags.target_tags.multi_output
+                tags.classifier_tags = deepcopy(last_step_tags.classifier_tags)
+                tags.regressor_tags = deepcopy(last_step_tags.regressor_tags)
+                tags.transformer_tags = deepcopy(last_step_tags.transformer_tags)
+        except (ValueError, AttributeError, TypeError):
+            # This happens when the `steps` is not a list of (name, estimator)
+            # tuples and `fit` is not called yet to validate the steps.
+            pass
+
+        return tags
+
 
 def _fit_resample_one(sampler, X, y, message_clsname="", message=None, params=None):
     with _print_elapsed_time(message_clsname, message):

imblearn/tests/test_common.py

@@ -9,10 +9,8 @@
 
 import numpy as np
 import pytest
-import sklearn
 from sklearn.exceptions import ConvergenceWarning
 from sklearn.utils._testing import ignore_warnings
-from sklearn.utils.fixes import parse_version
 
 from imblearn.over_sampling import RandomOverSampler
 from imblearn.under_sampling import RandomUnderSampler
@@ -32,14 +30,6 @@
 )
 from imblearn.utils.testing import all_estimators
 
-sklearn_version = parse_version(parse_version(sklearn.__version__).base_version)
-if sklearn_version >= parse_version("1.6"):
-    kwargs_parametrize_with_checks = {
-        "expected_failed_checks": _get_expected_failed_checks
-    }
-else:
-    kwargs_parametrize_with_checks = {}
-
 
 @pytest.mark.parametrize("name, Estimator", all_estimators())
 def test_all_estimator_no_base_class(name, Estimator):
@@ -49,7 +39,7 @@ def test_all_estimator_no_base_class(name, Estimator):
 
 
 @parametrize_with_checks_sklearn(
-    list(_tested_estimators()), **kwargs_parametrize_with_checks
+    list(_tested_estimators()), expected_failed_checks=_get_expected_failed_checks
 )
 def test_estimators_compatibility_sklearn(estimator, check, request):
     _set_checking_parameters(estimator)