[pymvpa] effect size (in lieu of zscore)

[Yaroslav Halchenko]

On Tue, 20 Dec 2011, Mike E. Klein wrote:
>    Does anyone know what could cause a negative shift in the searchlight
>    accuracy distribution for a 2-category Linear SVM classifier of single
>    subject? I'm seeing this in a couple of my subjects following a switch to
>    using zscore(dataset, chunks_attr='chunks', dtype='float32') …. so
>    zscoring against the mean and not against a baseline condition. (I'm
>    actually zscoring after removing the rest trials from the timeseries.)
>    While the vast majority of my subjects show a normal looking curve with a
>    peak right at 50%, a couple of subjects, while having a normal-esque
>    curve, show accuracy peaks at 40-45%. 

I think I have observed that as well, but blindly attributed it just to
a fact that peak might vary across subjects as well by chance... I should
have compared against not z-scored data I guess...

would you mind plotting/printing histograms of peaks across subjects
(how many?) with and without z-scoring?

>    Generally speaking, I'm seeing group data accuracy peaks in more-or-less
>    our hypothesized areas of the brain, but I'm still a bit worried that I've
>    done something wrong somewhere along the line. I can't figure out why I'm
>    seeing the negative peak (in a couple of subjects) or why using zscore
>    against rest has such a detrimental effect on my accuracies…

so it would be interesting to compare -- what if you make a scatter plot
(pylab.scatter) of such "problematic" subject searchlight performances
voxel by voxel (one axis performance with zscoring on the other --
without)?

NB I have cooked a little helper function to easily create such plots
and I haven't pulled it into PyMVPA yet -- I guess should do that
some time soon

-- 
=------------------------------------------------------------------=
Keep in touch                                     www.onerussian.com
Yaroslav Halchenko                 www.ohloh.net/accounts/yarikoptic