DOC add entry in changelog

doc/whats_new/0.13.rst

@@ -1,21 +1,35 @@
 .. _changes_0_13:
 
-Version 0.13.0 (Under development)
-==================================
+Version 0.13.0
+==============
 
-**TBD**
+**December 20, 2024**
 
 Changelog
 ---------
 
 Bug fixes
 .........
 
+- Fix `get_metadata_routing` in :class:`~imblearn.pipeline.Pipeline` such that one
+  can use a sampler with metadata routing.
+  :pr:`1115` by :user:`Guillaume Lemaitre <glemaitre>`.
+
 Compatibility
 .............
 
+- Compatibility with scikit-learn 1.6
+  :pr:`1109` by :user:`Guillaume Lemaitre <glemaitre>`.
+
 Deprecations
 ............
 
-Enhancements
-............
+- :class:`~imblearn.pipeline.Pipeline` now uses
+  :func:`~sklearn.utils.check_is_fitted` instead of
+  :func:`~sklearn.utils.check_fitted` to check if the pipeline is fitted. In 0.15, it
+  will raise an error instead of a warning.
+  :pr:`1109` by :user:`Guillaume Lemaitre <glemaitre>`.
+
+- `algorithm` parameter in :class:`~imblearn.ensemble.RUSBoostClassifier` is now
+  deprecated and will be removed in 0.14.
+  :pr:`1109` by :user:`Guillaume Lemaitre <glemaitre>`.

imblearn/ensemble/_weight_boosting.py

@@ -396,6 +396,7 @@ def _boost_discrete(self, iboost, X, y, sample_weight, random_state):
 
         return sample_weight, estimator_weight, estimator_error
 
+    # TODO(0.14): remove this method because algorithm is deprecated.
     def _boost(self, iboost, X, y, sample_weight, random_state):
         if self.algorithm != "deprecated":
             warnings.warn(