make features.Whitener compatible with sklearn whitener classes
from !60 (merged)
It comes down to implementing fit(), transform() and inverse_transform(), as well as set_params() and get_params() methods. So it looks like you could do it if we rename the methods we already have and drop the dependency on the names
parameter. However, since a lot of the whitener depends on names
there may be a massive rework to get everything to be API-compatible. Since there are no parameters set in init, set_params() and get_params() should be straightforward.
Here's the guide for reference: http://scikit-learn.org/stable/developers/contributing.html#rolling-your-own-estimator
When this is done, the sklearn classifiers should be allowed to use this whitener in addition to the other sklearn built-ins.
Edited by Reed Essick