We could do this by rejection sampling. But it could give very strange 
behaviors in corner cases, could lead to memory-leak-like properties of the 
object and I really don't see the benefits. It's just like bootstrap versus 
Jack knife. Jack knife has seldom practical value over bootstrap. The 
difference with what you are taking about is that it is ready to implement. 

Gaël

<div>-------- Original message --------</div><div>From: "Pagliari, Roberto" 
<[email protected]> </div><div>Date:14/08/2014  20:13  (GMT+01:00) 
</div><div>To: [email protected] </div><div>Subject: 
Re: [Scikit-learn-general] split function with non repeated sets </div><div>
</div>Here is a note from the documentation:

Note: contrary to other cross-validation strategies, random splits do not 
guarantee that all folds will be different, although this is still very likely 
for sizeable datasets.

So I guess it is not possible to enforce all sets to be different. 

Thank you


-----Original Message-----
From: Gael Varoquaux [mailto:[email protected]] 
Sent: Thursday, August 14, 2014 2:01 PM
To: [email protected]
Subject: Re: [Scikit-learn-general] split function with non repeated sets

> With k-fold you can shuffle the data, but let’s say you want to run 
> k-fold partition multiple times. Can you avoid repetition of the same 
> subset of samples?

We call that 'shuffle split'
http://scikit-learn.org/stable/modules/generated/sklearn.cross_validation.ShuffleSplit.html

Examples of use:
http://scikit-learn.org/stable/modules/cross_validation.html

Gaël

------------------------------------------------------------------------------
_______________________________________________
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
------------------------------------------------------------------------------
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to