Is it possible to use sklearn’s GroupKFold with the TreeFeatureReducer module? GroupKFold requires an array input groups while calling it’s split method but I can’t find a way to pass that argument while using this kind of cross validation in TreeFatureReducer.
Hey there,
If you look in the fit method of TreeFeatureReducer, you should be able to find the cv argument. I have not experimented with using grouped CV in TFR, so you may run into some problems! If you know what you’re doing though you might be able to edit the TFR to work with groupedCV… let me know how it goes.
Thanks,
Alex