Support full transform for OneHotEncoder about category_encoders HOT 10 CLOSED

Yeah, that makes sense. In get_dummies in one_hot we build up the encoded columns based on what is in the passed in X, but really ought to get the categories from the pre-trained OrdinalEncoder that got built in fit. I think that would solve this issue.

Is the assumption that all categories you want to encode are present in the dataset you pass to fit reasonable?

from category_encoders.

Yes, @wdm0006 your assumption is correct, all categories to encode should be passed in the dataset to fit.

from category_encoders.

Are you interested in working on a PR @justkunz? If not I'll try to work on this soon.

from category_encoders.

@wdm0006 I can get a quick PR up.

from category_encoders.

Fixed by #20

from category_encoders.

Will do a new release soon, would like to try to fix #18 beforehand though if possible.

from category_encoders.

@wdm0006 what's the status of the 1.2.4 release?

from category_encoders.

I need to do a docs update with some of the recent PRs, but will try to get it out this week.

from category_encoders.

@justkunz, just released it.

from category_encoders.

Thanks @wdm0006!

from category_encoders.