scikit-learn-contrib/metric-learn

transformer_ or components_ ?

Closed this issue · 1 comments

Currently we call the transformer matrix transformer_ for all algorithms. However in scikit-learn they mostly use components_ for the (quite) same object. I think we might have discussed this with @bellet and @nvauquie but I'm not sure at all, and don't remember what we decided
If it's better, it would be good to change this before the release to avoid a deprecation cycle

Indeed this seems more consistent to rename into components_ for MahalanobisMixin