Rankhot dev

[Rishoban]

This solution for enhancement Rank-Hot Encoder #199

[janmotl]

Nice.

Based on travis-ci, it is necessary to add 'RankHotEncoder' into category_encoders/__init__.py. But even after that, I got a few errors in test_rankhot on my computer. I suspect some versions of Pandas may cause troubles...

[Rishoban]

I will check that

On Mon, Nov 18, 2019 at 3:51 PM Jan Motl [email protected] wrote:

Nice.

Based on travis-ci, it is necessary to add 'RankHotEncoder' into category_encoders/init.py. But even after that, I got a few errors in test_rankhot on my computer. I suspect some versions of Pandas may cause troubles...

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/scikit-learn-contrib/categorical-encoding/pull/222?email_source=notifications&email_token=AG5GGUPGXQO3QGLWMRA3FV3QUJUD3A5CNFSM4JOQMW2KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEEJ6HOY#issuecomment-554951611, or unsubscribe https://github.com/notifications/unsubscribe-auth/AG5GGUOF6W5RPAHRZDCNDUDQUJUD3ANCNFSM4JOQMW2A .

[wdm0006]

@PaulWestenthanner are you interested in taking a look at this PR? It needs to be rebuilt with the newer CI to see if the issues are resolved and have some merge conflicts cleaned up.

[PaulWestenthanner]

Hi @Rishoban I'm very sorry for being so late to come back to this PR. I've done a lot of work on it and it now fits nicely into the existing structure (features like return_nan as handle_missing or handle_unknown). I also made all the tests work and could simplify the code in some places.
Please have a look, (or anybody else). Feedback is always appreciated