[pymvpa] Random choosing samples within each fold

[gal star]

Hello Experts,
Is there a way to randomly choose samples for each fold using pymvpa API?

So far, I have managed to use NFoldPartitioner to split dataset
into folds only by using 'chunks' (that i've pre-set in attributes.txt
file).

I want to see if i mix all samples -
whether the classifier performs better on my dataset.

There are relevant flags 'selection_strategy' and 'count' for the
NFoldPartitioner -
but when i use those, I get the following error:
"Got empty training dataset from splitting in TransferMeasure. unique
values of input split attributes are: [2]"

Could anyone help?

Thanks,
Gal Sabina Star
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/attachments/20141201/c6a8d9b0/attachment.html>