Am 10.12.2012 15:55, schrieb Conrad Lee:
> Say I have some arbitrary score function my_func() that I want 
> GridSearchCV to optimize.  Let's say I'm doing classification and not 
> regression.
>
> GridSearchCV will use a classifier's predict() function to make 
> predictions, and pass those predictions into the my_func, along with 
> the true values.  But what if my_func needs the output of 
> predict_proba to arrive at its score?  Is there any way I can get 
> GridSearchCV use do predict_proba with my_func?
>
I have been working on that for quite a while. After many attempts (see 
https://github.com/scikit-learn/scikit-learn/pull/1014 and 
https://github.com/scikit-learn/scikit-learn/pull/1198), and some 
discussions, we settled an interface to do that.

My current PR is https://github.com/scikit-learn/scikit-learn/pull/1381.

It is pretty high up in my priority list, but currently far from mergable.
Not sure if I should try to get it into the release or not.
Will most definitely work on it next weekend.

Hope that helps.
Andy

------------------------------------------------------------------------------
LogMeIn Rescue: Anywhere, Anytime Remote support for IT. Free Trial
Remotely access PCs and mobile devices and provide instant support
Improve your efficiency, and focus on delivering more value-add services
Discover what IT Professionals Know. Rescue delivers
http://p.sf.net/sfu/logmein_12329d2d
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to