[pymvpa] Linear SVM classifiers

[Mike E. Klein]

Hi all,

Could anyone tell me if there's a tangible difference between using the
LinearNuSVMC() and LinearCSVMC() classifiers? Also, in order to have the
classifier choose its own best "fit," is it best practice to leave the area
between the brackets blank, or to but in a value of C=-1 ? I think I've
heard of doing either/both, but may be misremembering. I've posted before
about searchlight analyses not running with LinearCSVMC unless specific a
C-parameter is input, so I'm wondering the best choice to make.

Best,
Mike
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/attachments/20120104/440cfcf5/attachment.html>