[pymvpa] Train and test on different classes from a dataset

[Anthony Sali]

Dear all,

I am attempting to train a classifier on trials of classes A and C and test
the classifier on trials of classes B and D from a left out run (where B
corresponds to A and D corresponds to C).  I found an earlier message board
post with a very helpful script to perform this kind of partitioning (
http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/2013q1/002372.html).


I'm new to python and pymvpa and consequently I am stuck regarding how to
implement this partitioning into a searchlight analysis.  I thought that I
would be able to use the chain node created in the script when defining the
cross validation with:

cvte = mv.CrossValidation(clf, chain)
s1 = mv.sphere_searchlight(cvte, radius=3, postproc=mv.mean_sample(),
nproc=1)
res = s1(evds)

However, I get an error of:"TypeError: argument of type 'NoneType' is not
iterable."

I'd appreciate any suggestions on where I'm going wrong.

Thanks!
Anthony
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/attachments/20150528/9fe56dda/attachment.html>