[pymvpa] Loading normalized data with 'get_model_bold_dataset' fails

lydia.hellrung at econ.uzh.ch lydia.hellrung at econ.uzh.ch
Thu Jun 1 12:22:48 UTC 2017



Dear pymvpa-Team,

I'm new to your toolbox (running version 2.6.0) and I got stucked while
data loading from my own dataset and right now I really don't know what the
issue could be and what to look up.
So, I sorted data analog to openfmri structure and paths are all fine since
I can load one data file but not the other.
My loading procedure looks as follows and it works fine for a different
flavor (here, I try to load normalized data, but it works with the
unnormalized ones from the same subject):

of = OpenFMRIDataset(datapath)
print of.get_task_descriptions()
conditions = of.get_model_conditions(1)
print conditions
print [c['name'] for c in conditions]
ds = of.get_model_bold_dataset(
   model_id=1,
   subj_id=2,
   flavor='iwrbadvols',
   mask=opj(datapath, 'sub002', 'masks', 'test', 'brain.nii.gz'),
   modelfx=fit_event_hrf_model,
   time_attr='time_coords', condition_attr='condition')
print "Data loaded"

When I try to load the normalized data, I always end up in the numpy file
"function_base.py" in line 76 with "num = int(num)" and the issue is that
"num" in this case has "Inf" as value.

The error message says exactly:
" The debugged program raised the exception OverflowError (cannot convert
float infinity to integer)"
(plus file reference to function_base.py and line 76)

So, data are from same subject, look fine if I inspect them, just the
normalized ones of course are resampled to the normalization template (so
there are 3x more data with 770MB just this file).
And I can't see anything how to set/influence this variable.

I really can't figure out where I have to look now to fix this - could
there be something wrong with the data or is the file size just too big? Do
I have to use another option for the function call?

Thanks for any advice
Lydia

P.S: I'm not sure if this is relevant: the toolbox is running on a virtual
machine on a middle performing computer (at least the VM is using 5 GB RAM
and 4 processors)

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/pkg-exppsy-pymvpa/attachments/20170601/ce455531/attachment.html>


More information about the Pkg-ExpPsy-PyMVPA mailing list