This issue tracks the status of follow-ups of #18514, i.e. the implementation of SLEP10.

According to N_FEATURES_IN_AFTER_FIT_MODULES_TO_IGNORE in test_common.py, as of 96a96f1, modules to be n_feature_in_-ified are:

• calibration Test and doc for n_features_in_ for sklearn.calibration #19555
• compose TST Add TransformedTargetRegressor to test_meta_estimators_delegate_data_validation #20175
• covariance ENH Checks n_features_in_ in covariance #19341
• discriminant_analysis ENH Checks n_features_in_ in discriminant_analysis #19342
• ensemble ENH Adds n_features_in_ to ensemble module #19326
• feature_extraction MNT add n_features_in_ through the feature_extraction module #20180
• feature_selection ENH Adds n_features_in_ checking to feature_selection #19344
• isotonic TST Enables isotonic and manifold in test_check_n_features_in_after_fitting #19539
• manifold TST Enables isotonic and manifold in test_check_n_features_in_after_fitting #19539
• mixture ENH Checks n_features_in_ after fitting in mixture #19540
• model_selection TST common tests for n_features_in_ for model_selection module #20204
• multiclass MNT n_features_in through the multiclass module #20193
• multioutput ENH Adds n_features_in_ checking to multioutput #19692
• naive_bayes ENH Adds n_features_in_ to naive_bayes #19485
• pipeline TST check n_features_in_ in pipeline module #20192
• random_projection ENH Checks n_features_in_ after fitting in random_projection #19541

Track documentation status of n_features_in_ , see N_FEATURES_MODULES_TO_IGNOREin test_docstring_parameters.py, start was #19351. Note that existing alternatives like n_features_ attributes have to be properly deprecated:

• calibration Test and doc for n_features_in_ for sklearn.calibration #19555
• cluster DOC Document n_features_in_ in cluster #20228
• cross_decomposition DOC Documents n_features_in_ in cross_decomposition #19351
• compose TST Add TransformedTargetRegressor to test_meta_estimators_delegate_data_validation #20175
• covariance DOC add n_features_in_ in the documentation #20236
• decomposition DOC add n_features_in_ in the documentation #20236
• discriminant_analysis DOC add n_features_in_ in the documentation #20236
• dummy DOC add n_features_in_ in the documentation #20236
• ensemble DOC add n_features_in_ in the documentation #20236
• feature_extraction
• feature_selection DOC add n_features_in_ in the documentation #20236
• gaussian_process DOC add n_features_in_ in the documentation #20236
• impute DOC add n_features_in_ in the documentation #20236
• isotonic
• kernel_approximation DOC add n_features_in_ in the documentation #20236
• kernel_ridge DOC add n_features_in_ in the documentation #20236
• linear_model DOC add n_features_in_ in the documentation #20236
• manifold DOC add n_features_in_ in the documentation #20236
• mixture ENH Checks n_features_in_ after fitting in mixture #19540
• model_selection TST common tests for n_features_in_ for model_selection module #20204
• multiclass MNT n_features_in through the multiclass module #20193
• multioutput ENH Adds n_features_in_ checking to multioutput #19692
• naive_bayes DOC add n_features_in_ in the documentation #20236
• neighbors DOC add n_features_in_ in the documentation #20236
• neural_network DOC add n_features_in_ in the documentation #20236
• pipeline TST check n_features_in_ in pipeline module #20192
• preprocessing DOC add n_features_in_ in the documentation #20236
• random_projection
• semi_supervised DOC add n_features_in_ in the documentation #20236
• svm DOC add n_features_in_ in the documentation #20236
• tree DOC add n_features_in_ in the documentation #20236

Note: for meta-estimators it is better to delegate feature consistency validation to the inner base estimators.