[pymvpa] Pairwise classification

Jonas Kubilius jonas.kubilius at ppw.kuleuven.be
Wed Jan 11 13:54:47 UTC 2012

> I guess you should generate the pairs yourself.

Maybe I should phrase it differently then: why isn't pairwise
classification the default? Is it somehow inferior? Has some bias? I've
never used multiclass classification so I am not sure about the differences
between the two approaches.

> But pay attention that estimating the accuracy using "all" possible pairs
> of the
> test set incurs in an optimisitc bias. The pairs are not independent since
> they sometimes share one element. Having a test set of non-independent
> examples leads to optimistic bias. In order not to have that bias you
> should
> use instead the subset of non-overlapping pairs.

Thanks! I'm well aware of non-indepence issues.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/attachments/20120111/277ab606/attachment.html>

More information about the Pkg-ExpPsy-PyMVPA mailing list