The usage of max_features="auto" as parameter value in RandomForestRegressors and ExtraTreesRegressors is depreciated in scikit-learn 1.1 and will be removed in 1.3. The default value for max_features also changed from from "auto" to "sqrt" in version 1.1.

In analysis/feature_scoring.py this function is used:

The full warning from the CI:

test_get_rf_feature_scores (test.test_analysis.test_feature_scoring.FeatureScoringTestCase) ... /opt/hostedtoolcache/Python/3.9.12/x64/lib/python3.9/site-packages/sklearn/ensemble/_forest.py:427: FutureWarning: max_features='auto' has been deprecated in 1.1 and will be removed in 1.3. To keep the past behaviour, explicitly set max_features='sqrt' or remove this parameter as it is also the default value for RandomForestClassifiers and ExtraTreesClassifiers.

See scikit-learn/scikit-learn#20111 for the motivation and discussion behind the depreciation.

@quaquel What new value is most appropriate?