2012/10/24 Alexandre Gramfort <[email protected]>:
> hi folks,
>
> I apologize if this was already a discarded option above but I read quickly.
>
> how about putting the logic in cross_val_score and GridSearchCV and just
> "inspect" if the score_func to see it expects a parameter called "estimator"?
> that would then mean calling the score func with (estimator, X) for 
> unsupervised
> and (estimator, X, y) for supervised.
>
> the existing code would still work and for advanced use cases it could fly to.

Arguments inspection would be a good way to deal with backward compat
indeed. However I would still mark the non-estimator API deprecated
with a warning with maybe +4 releases (1 year deprecation) to get rid
of arg introspection later and always require the __call__(estimator,
X, y=None) API in the long run.

-- 
Olivier
http://twitter.com/ogrisel - http://github.com/ogrisel

------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_sfd2d_oct
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to