Currently the `getFeaturesCols` will return the fold column and the weights column. We should update this method so that it only returns columns that are used to train on.
It should provably print warning, letting the user know about this behavior.
Also, from data science point of view, , can you verify that excluding these 2 columns for scoring does make sense? I don't have the DS background. Thanks!
hi yes I think it makes sense to exclude these two columns from a DS perspective, because the fold column and weights column shouldn't be used as features during training, and the fold column in particular shouldn't be required during scoring. hope this helps!
Cool! thank you . This change will overlap with big API clean up introduced by ( will go into major release)