Changelog

Bug fixes

• Fixget_metadata_routing inPipeline such that one can use a sampler with metadata routing.#1115 by@glemaitre.

Compatibility

• Compatibility with scikit-learn 1.6.#1109 by@glemaitre.

Deprecations

• Pipeline now usescheck_is_fitted. In 0.15, it will raise an error instead of a warning.#1109 by@glemaitre.

• algorithm parameter inRUSBoostClassifier is now deprecated and will be removed in 0.14.#1109 by@glemaitre.

Contributors

Changelog

Compatibility

• Compatibility with NumPy 2.0+#1097 byGuillaume Lemaitre.

Changelog

Compatibility

• Compatibility with scikit-learn 1.5#1074 and#1084 byGuillaume Lemaitre.

Changelog

Bug fixes

• Fix the way we check for a specific Python version in the test suite.#1075 byGuillaume Lemaitre.

Changelog

Bug fixes

• Fix a bug inInstanceHardnessThreshold where estimator could not be a Pipeline object.#1049 byGonenc Mogol.

Compatibility

• Do not use distutils in tests due to deprecation.#1065 byMichael R. Crusoe.
• Fix the scikit-learn import in tests to be compatible with version 1.4.1.post1.#1073 byGuillaume Lemaitre.
• Fix test to be compatible with Python 3.13.#1073 byGuillaume Lemaitre.

Changelog

Bug fixes

• Fix a bug inSMOTENC where the entries of the one-hot encoding should be divided by sqrt(2) and not 2, taking into account that they are plugged into an Euclidean distance computation.#1014 byGuillaume Lemaitre.
• Raise an informative error message when all support vectors are tagged as noise inSVMSMOTE.#1016 byGuillaume Lemaitre.
• Fix a bug inSMOTENC where the median of standard deviation of the continuous features was only computed on the minority class. Now, we are computing this statistic for each class that is up-sampled.#1015 byGuillaume Lemaitre.
• Fix a bug inSMOTENC such that the case where the median of standard deviation of the continuous features is null is handled in the multiclass case as well.#1015 byGuillaume Lemaitre.
• Fix a bug inBorderlineSMOTE version 2 where samples should be generated from the whole dataset and not only from the minority class.#1023 byGuillaume Lemaitre.
• Fix a bug inNeighbourhoodCleaningRule where the kind_sel="all" was not working as explained in the literature.#1012 byGuillaume Lemaitre.
• Fix a bug inNeighbourhoodCleaningRule where the threshold_cleaning ratio was multiplied on the total number of samples instead of the number of samples in the minority class.#1012 byGuillaume Lemaitre.
• Fix a bug inRandomUnderSampler andRandomOverSampler where a column containing only NaT was not handled correctly.#1059 byGuillaume Lemaitre.

Compatibility

• BalancedRandomForestClassifier now support missing values and monotonic constraints if scikit-learn >= 1.4 is installed.
• Pipeline support metadata routing if scikit-learn >= 1.4 is installed.
• Compatibility with scikit-learn 1.4.#1058 byGuillaume Lemaitre.

Deprecations

• Deprecate estimator_ argument in favor of estimators_ for the classesCondensedNearestNeighbour andOneSidedSelection. estimator_ will be removed in 0.14.#1011 byGuillaume Lemaitre.
• Deprecate kind_sel in#1012 byGuillaume Lemaitre.

Enhancements

• Allows to output dataframe with sparse format if provided as input.#1059 byts2095.

Changelog

Bug fixes

• Fix a bug inclassification_report_imbalanced where the parametertarget_names was not taken into account whenoutput_dict=True.#989 byAYY7.

• SMOTENC now handles mix types of data type such as bool andpd.CategoricalDtype by delegating the conversion to scikit-learn encoder.#1002 byGuillaume Lemaitre.

• Handle sparse matrices inSMOTEN and raise a warning since it requires a conversion to dense matrices.#1003 byGuillaume Lemaitre.

• Remove spurious warning raised when minority class get over-sampled more than the number of sample in the majority class.#1007 byGuillaume Lemaitre.

Compatibility

• Maintenance release for being compatible with scikit-learn >= 1.3.0.#999 byGuillaume Lemaitre.

Deprecation

• The fitted attributeohe_ inSMOTENC is deprecated and will be removed in version 0.13. Usecategorical_encoder_ instead.#1000 byGuillaume Lemaitre.

• The default of the parameterssampling_strategy and replacement will change inBalancedRandomForestClassifier to follow the implementation of the original paper. This changes will take effect in version 0.13.#1006 byGuillaume Lemaitre.

Enhancements

• SMOTENC now accepts a parametercategorical_encoder allowing to specify aOneHotEncoder with custom parameters.#1000 byGuillaume Lemaitre.

• SMOTEN now accepts a parametercategorical_encoder allowing to specify aOrdinalEncoder with custom parameters. A new fitted parametercategorical_encoder_ is exposed to access the fitted encoder.#1001 byGuillaume Lemaitre.

• RandomUnderSampler andRandomOverSampler (whenshrinkage is notNone) now accept any data types and will not attempt any data conversion.#1004 byGuillaume Lemaitre.

• SMOTENC now support passing array-like ofstr when passing thecategorical_features parameter.#1008 by :userGuillaume Lemaitre <glemaitre>.

• SMOTENC now support automatic categorical inference whencategorical_features is set to"auto".#1009 by :userGuillaume Lemaitre <glemaitre>.

Changelog

Bug fixes

• Fix a regression in over-sampler where the stringminority was rejected as an unvalid sampling strategy.#964 by Prakhyath07.

Changelog

Bug fixes

• Make sure that Substitution is working withpython -OO that replacesdoc by None.#953 buGuillaume Lemaitre.

Compatibility

• Maintenance release for being compatible with scikit-learn >= 1.0.2.#946,#947,#949 byGuillaume Lemaitre.
• Add support for automatic parameters validation as in scikit-learn >= 1.2.#955 byGuillaume Lemaitre.
• Add support forfeature_names_in_ as well asget_feature_names_out for all samplers.#959 byGuillaume Lemaitre.

Deprecation

• The parametern_jobs has been deprecated from the classesADASYN,BorderlineSMOTE,SMOTE,SMOTENC,SMOTEN, andSVMSMOTE. Instead, pass a nearest neighbors estimator where n_jobs is set.#887 byGuillaume Lemaitre.
• The parameterbase_estimator is deprecated and will be removed in version 0.12. It is impacted the following classes:BalancedBaggingClassifier,EasyEnsembleClassifier,RUSBoostClassifier.#946 byGuillaume Lemaitre.

Enhancements

• Add support to accept compatible NearestNeighbors objects by only duck-typing. For instance, it allows to accept cuML instances.#858 byNV-jpt andGuillaume Lemaitre.

Compatibility with scikit-learn 1.1.0