Dear Developers,
Currently one can give sklearn.feature_selection.SequentialFeatureSelector
parameter 'n_features_to_select'
to get desired number of features.
It would be desired, that instead of giving a fixed number of 'n_features_to_select' one could use it as an upper limit and select features only up to point where the score improves (so the number of selected features could be smaller than 'n_features_to_select').
Terveisin, Markus