sklearn.feature_selection.SequentialFeatureSelector Select features as long as score gets better.

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

[NicolasHug]

Yes this sounds good.

An alternative that I slightly prefer would be to directly introduce tol=0 and instead do the deprecation on n_features_to_select:

• set n_features_to_select to ‘warn’ indicating that None is deprecated in favor of ‘auto’
• Switch the default to ‘auto’ in 2 versions and remove None.

The reason the default is currently None is for consistency with RFE, but ‘auto’ is a more accurate name. Since we have to deprecate something anyway, we might as well introduce a better name in the process.

I’m fine with either strategy anyway, so feel free to implement the one you like best.

[ogrisel]

@NicolasHug do you agree with the plan above? If so we can implement it as part of #20145.