Version 0.7.0#
June 9, 2020
Changelog#
Maintenance#
Ensure that
imblearn.pipeline.Pipelineis working whenmemoryis activated andjoblib==0.11. #687 by Christos Aridas.Refactor common test to use the dev tools from
scikit-learn0.23. #710 by Guillaume Lemaitre.Remove
FutureWarningissued byscikit-learn0.23. #710 by Guillaume Lemaitre.Impose keywords only argument as in
scikit-learn. #721 by Guillaume Lemaitre.
Changed models#
The following models might give some different results due to changes:
Bug fixes#
Change the default value
min_samples_leafto be consistent with scikit-learn. #711 by zerolfx.Fix a bug due to change in
scikit-learn0.23 inimblearn.metrics.make_index_balanced_accuracy. The function was unusable. #710 by Guillaume Lemaitre.Raise a proper error message when only numerical or categorical features are given in
imblearn.over_sampling.SMOTENC. #720 by Guillaume Lemaitre.Fix a bug when the median of the standard deviation is null in
imblearn.over_sampling.SMOTENC. #675 by bganglia.
Enhancements#
The classifier implemented in imbalanced-learn,
imblearn.ensemble.BalancedBaggingClassifier,imblearn.ensemble.BalancedRandomForestClassifier,imblearn.ensemble.EasyEnsembleClassifier, andimblearn.ensemble.RUSBoostClassifier, acceptsampling_strategywith the same key than inywithout the need of encodingyin advance. #718 by Guillaume Lemaitre.Lazy import
kerasmodule when importingimblearn.keras#719 by Guillaume Lemaitre.
Deprecation#
Deprecation of the parameters
n_jobsinimblearn.under_sampling.ClusterCentroidssince it was used bysklearn.cluster.KMeanswhich deprecated it. #710 by Guillaume Lemaitre.Deprecation of passing keyword argument by position similarly to
scikit-learn. #721 by Guillaume lemaitre.