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

Michael Hanke mih at debian.org
Fri Jan 18 13:49:24 UTC 2013


On Wed, Jan 16, 2013 at 09:56:07PM +0100, Jan Derrfuss wrote:
> It would be great if you could give me another hint!

Well, the hint with the Sifter wasn't really useful (it doesn't do what
you want, as far as I understand it now) -- apologies!

Attached is a script that should demo how to actually achieve what you want.
Just run it on a machine with PyMVPA and Numpy installed. It basically ammends
the NFoldPartitioner with some advanced magic ... and voila ... ;-)

(Comments are inside)

Let me know if you need help merging this code into your script.



Michael Hanke
-------------- next part --------------
A non-text attachment was scrubbed...
Name: sifter.py
Type: text/x-python
Size: 3100 bytes
Desc: not available
URL: <http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/attachments/20130118/c4c4acd1/attachment.py>

More information about the Pkg-ExpPsy-PyMVPA mailing list