[pymvpa] SVM + RFE

[Yaroslav Halchenko]

On Mon, 27 Jan 2014, Arman Eshaghi wrote:
>    Summary for chunks across targets
>      chunks mean std min max #targets
>        0     21   3   18  24     2
>        1    27.5 2.5  25  30     2

so that is the reason -- you have only 2 chunks.

For such RFE setup you would need at least 3 since it tries to estimate
(without picking) the stopping point by using cross-validation within
the training set.  if training set contains only 1 chunk -- it can't
split more...   

to overcome this you could partition each chunk more, e.g. assigned
.sa['subchunks'] and specify to use that one instead of chunks 
in your splitter using in RFE definition.

-- 
Yaroslav O. Halchenko, Ph.D.
http://neuro.debian.net http://www.pymvpa.org http://www.fail2ban.org
Senior Research Associate,     Psychological and Brain Sciences Dept.
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