On Apr 17, 2012, at 15:53 , Alexandre Gramfort wrote:

>> I think just moving from a train set to a test set would be problematic for 
>> small n_samples.
> 
> what do you suggest?
> 

I agree with your scale_C=None suggestion because it would (in theory) force 
the user to become aware of what the setting means. For this to work there are 
a couple of points I'd like to raise:

1. To make progress in the fight with the windmills, the docs should make it 
clear that the libsvm default is a bad idea,  like with a toy example that 
would break.

2. I fear that dismissive users will start destructive word-of-mouth: "sklearn 
svms don't work <<correctly>> unless you set scale_C=True", and users would end 
up setting scale_C to True without reading the docs.

I think this needs more than just clarification in the docs. I think we need 
some sort of "Release notes" or "News" highlight on the front page. A blog post 
maybe, that should be made viral in some way. SVMs are probably one of our most 
used modules.

Vlad


> Alex
> 
> ------------------------------------------------------------------------------
> Better than sec? Nothing is better than sec when it comes to
> monitoring Big Data applications. Try Boundary one-second 
> resolution app monitoring today. Free.
> http://p.sf.net/sfu/Boundary-dev2dev
> _______________________________________________
> Scikit-learn-general mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/scikit-learn-general


------------------------------------------------------------------------------
Better than sec? Nothing is better than sec when it comes to
monitoring Big Data applications. Try Boundary one-second 
resolution app monitoring today. Free.
http://p.sf.net/sfu/Boundary-dev2dev
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to