I agree, this sounds like a very useful addition. Why I was a bit hesitant
about adding it to the initial implementation was that I was/am not sure what
happens if, for example, in cases such as
cross_val_score w. n_jobs > -1
VotingClassifier w. n_jobs > -1
RandomForest w. n_jobs > -1
Using the code from cross_val_score, would this safely execute? I think if the
outer functions would use up all of the CPUs already, the inner functions would
basically run on a single CPU without throwing errors!? I could try adding
multiprocessing to the VotingClassifier, I think that’s very useful in certain
situations where the estimators inside don’t use multiprocessing themselves. Or
I could leave this feature for you to implement if you like ;) ??
> On Nov 11, 2015, at 8:27 PM, Scott Turner <[email protected]> wrote:
>
> On Wed, Nov 11, 2015 at 6:18 PM,
> <[email protected]
> <mailto:[email protected]>> wrote:
> @Scott: See https://github.com/scikit-learn/scikit-learn/pull/5794
> <https://github.com/scikit-learn/scikit-learn/pull/5794> for checking the
> VotingClassifier
>
> It has also occurred to me that VotingClassifier ought to be capable of using
> multiple CPUs to run the base classifiers, ala cross_val_score. The code
> there is fairly easy to copy.
>
> -- Scott
> ------------------------------------------------------------------------------
> _______________________________________________
> Scikit-learn-general mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
------------------------------------------------------------------------------
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
