Add leave-one-out encoding about category_encoders HOT 6 CLOSED

I'll tackle this one if no one else is on it yet.

from category_encoders.

Awesome, let me know if you have any questions along the way!

from category_encoders.

I want to try this, but i wonder what should i do when i meet a never seen category?
Encoder it as the mean of y in whole train data or something else?

from category_encoders.

@hbghhy awesome! In one-hot, you'll see an example of how we've handled this in the past. There we have options of raising exceptions, ignoring the samples, or imputing something to indicate 'unknown'. For this, imputing the mean of whole training data may be a good approach for 'unknown'.

from category_encoders.

@wdm0006 ,I just pull a new request about it. You can see is there something wrong.

from category_encoders.

Fixed!

from category_encoders.