On 03/12/2013 02:05 PM, Joel Nothman wrote:
> Andreas wrote:
>
>> In the meantime, did you have a look at
>> https://github.com/scikit-learn/scikit-learn/pull/1742?
> No, I hadn't, but now I've merged that (not a trivial merge), and
> changed a couple of things a little.
>
> https://github.com/jnothman/scikit-learn/tree/grid_search_more_info 
> (6e71aeaf8c)
>
> Your output was a list of name-spaces. My output is a name-space of
> lists. I think therefore what we really want is a numpy structured
> array, which can be indexed either by field or by index (still
> enabling vectorised calculations), and whose field names are fixed as
> in a namedtuple.
Sorry for being terse, I should be working.
We really can not break backward compatibility.
One possibility is to have the old ``grid_scores_`` be the same as
it was before and make ``cv_scores_`` have the new structure.

I agree with you that a structure of lists is nicer.
Whenever I wanted to plot, I had to do
[score.property for score in grid.cv_scores_].
Cheers,
Andy


------------------------------------------------------------------------------
Symantec Endpoint Protection 12 positioned as A LEADER in The Forrester  
Wave(TM): Endpoint Security, Q1 2013 and "remains a good choice" in the  
endpoint security space. For insight on selecting the right partner to 
tackle endpoint security challenges, access the full report. 
http://p.sf.net/sfu/symantec-dev2dev
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to