[pymvpa] ERNiftiDataset with different event durations

Matthias Ekman Matthias.Ekman at nf.mpg.de
Fri Jul 17 12:59:56 UTC 2009

Yaroslav Halchenko wrote:
>> and the accuracy substantially decreased (may be this effect could
>> be "undone" with correlation stability analysis before
>> classification).
> So, you did binary classification, right?

yes binary classification, using LinearNuSVMC(nu=0.5, probability=0)
> was it full brain or some ROI? what # of voxels?
full brain with 34000 features
> TR was 2 or 3 seconds so changing duration from 6 to 7 added another
> sample into 'spatio-temporal' pattern, right? 
right. TR is 2 sec. so one-(half) samlpe (+features) is added
> Depending on the
> classifier, the effect of substantial increase of number of feature
> could have confused it, unless you've used some classifier with feature
> selection... so what classifier was it (family/parameters/etc
I did some classifications with features selection (stability analysis; 
# features = 5000) and the results with event duration of 7 sec. are 
still worth compared to the results with (true) event duration of 6 sec. 
I'll to some further classifications (with fewer voxels) and will report.

Best regards,

More information about the Pkg-ExpPsy-PyMVPA mailing list