Sorry about that oversight in the design! A common test to catch those sorts of inconsistencies would be useful.
The biggest problem is that KernelDensity is not fundamentally a classifier, regressor, or transformer, but a density estimator. When I initially did the KDE pull request, I floated the idea of defining a standard density estimator interface which would be a mixin like ``ClassifierMixin`` and ``RegressorMixin``. People involved at the time didn't seem particularly excited about that, and there is the added issue that GMM has some (IMO) counterintuitive API choices on the density estimator side which would make such a change difficult. I still would contend that settling-on and defining a density estimator API would make the library much more usable for estimators like KernelDensity, GMM, and other generative unsupervised methods. My two cents, Jake Jake VanderPlas Director of Research – Physical Sciences eScience Institute, University of Washington http://www.vanderplas.com On Wed, Nov 5, 2014 at 8:29 AM, Andy <t3k...@gmail.com> wrote: > Fix here: > https://github.com/scikit-learn/scikit-learn/pull/3826 > > > ------------------------------------------------------------------------------ > _______________________________________________ > Scikit-learn-general mailing list > Scikit-learn-general@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/scikit-learn-general >
------------------------------------------------------------------------------
_______________________________________________ Scikit-learn-general mailing list Scikit-learn-general@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
