[pymvpa] RSA correlation > 1 ?

[David Soto]

Hi, I am computing the similarity between the activation maps of 2
conditions within a frontoparietal mask using the code below,
, and, for each individual subject. I am getting correlation values higher
than 1, see output list at the bottom, however I understand these should
range between 0 (identical) to 1 (uncorrelated)?
Any advise is appreciated
thanks!
david


ps=[0, 1,2,3,4,5,6,7,8,9,10,11,12,13, 14,15,16,17,18]

resultado=[]
from mvpa2.mappers.detrend import poly_detrend

for subj in ps:
    ds = fmri_dataset(samples=os.path.join(datapath,
's.i.pred%d.nii.gz')%(int(subj)),
                      targets=attr.targets, chunks=attr.chunks,
                      mask=os.path.join(datapathmask, 'mask.nii.gz'),#based
on FEAT analyses
                      add_fa={'cluster_searchpred': os.path.join(datapath1,
'cluster_searchpred.nii.gz')})

    from mvpa2.measures import rsa
    rsa_xsubj = rsa.PDist(square=True)
    res = rsa_xsubj(ds)
    resul=res.S[0,1]
    resultado.append(res.S[0,1])
---->
0.60377539435397509,
 0.61374525419685622,
 1.1142951914849382,
 0.97604011146884972,
 0.49202801554753195,
 0.77693568183436723,
 1.1792308174154826,
 0.79977620949819672,
 0.82386196165868597,
 0.73430209533220103,
 0.75513093298950229,
 1.0864677653799166,
 1.0623285688583259,
 0.97273572366115335,
 0.96046313836217112,
 0.2798419853674019,
 0.85927153773083831,
 1.6251078728068848,
 0.44645101975074164,
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/attachments/20160613/5b5a31fc/attachment.html>