[pymvpa] multiclass hit rate conflicts with two classes hit rate results

[Vadim Axel]

Hi,

I have 5 classes design. When I run multiclass SVM I am getting ~2 or 3
categories beyond chance results. However, when I run all pairs two classes
classification, the results are beyond chance level in all the
classifications (0.55-0.65 % hit rate).  As far as I understand, this is a
possible scenario with the major voting scheme of multiclass classification
of all possible pairs. My question is how should one interpret such results?
Doubtfully, I would prefer to get beyond chance level in mutliclass as well.
However, since in pair classification I get beyond chance level, can it
serve by itself  an evidence for neural encoding scheme I try to decode?

Thanks for help,
Vadim
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/attachments/20091031/61b85477/attachment.htm>