[pymvpa] t-maps in PyMVPA

Yaroslav Halchenko debian at onerussian.com
Fri Mar 11 13:31:51 UTC 2016


On Fri, 11 Mar 2016, Salim Al-wasity wrote:

>    Dears
>    I used the MVPA package of Brainvoyager for conditions classification. The
>    SVM classifier was trained and tested using t-maps values extracted after
>    a GLM was fitted to my data.
>    Is there any way to replicate this analysis and get the t-values using
>    PyMVPA?

sure

keywords:

fmri_dataset, 
CrossValidation, 
LinearCSVMC(C=1) (I guess you used that one),
ttest_1samp

I want to bring to your attention though a recent nice exploration of
validity of t-tests (or bootstrap/permutation) on accuracy (or any other
information metric) for stating typicality of the discovered effect at
population level:
http://arxiv.org/abs/1512.00810
here is my quick replication of the initial simulation
http://nbviewer.jupyter.org/url/www.onerussian.com/tmp/test_ttest.ipynb
So if in your design you could use cross-subject cross-validation you
might be much better ;)
-- 
Yaroslav O. Halchenko
Center for Open Neuroscience     http://centerforopenneuroscience.org
Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755
Phone: +1 (603) 646-9834                       Fax: +1 (603) 646-1419
WWW:   http://www.linkedin.com/in/yarik        



More information about the Pkg-ExpPsy-PyMVPA mailing list