[pymvpa] Tutorial data: did something change?
shanusmagnus at gmail.com
Wed Jul 2 06:05:26 UTC 2014
Hi Yaroslav -
Thanks for taking the time to respond. A point of clarification, if you
Original tutorial data is still out there from the original location and
> it is the one which actually should be used to go through the tutorial
> -- nothing really changed in this regard.
I'm looking through the directory where macports installed PyMVPA. In that
subtree, there's only one occurrence of the mask_vt.nii.gz file mentioned
in the tutorial (for instance, here:
http://www.pymvpa.org/tutorial_mappers.html) which is in the new
tutorial_data_25mm/data/ subdir. The confusing thing is that this mask has
only a single non-zero voxel; which means that when you go through this
section of the tutorial (on the same page as linked above):
>>> fds = fmri_dataset(samples=os.path.join(datapath, 'bold.nii.gz'),... targets=attr.targets, chunks=attr.chunks,... mask=os.path.join(datapath, 'mask_vt.nii.gz'))>>> fds.shape(1452, 577)>>> print fds.sa<SampleAttributesCollection: chunks,targets,time_coords,time_indices>
the fds.shape actually reports (1452, 1).
If I'm interpreting your reply correctly, the old tutorial data, with a
mask_vt.nii.gz that includes 577 features, is supposed to still be
distributed with PyMVPA 2.31, correct? Let me check the git repo ... okay,
a fresh checkout of the git repo is the same -- the only copy of
mask_vt.nii.gz is in tutorial_data_25mm/data/ and consists of a single
giant non-zero voxel.
So it seems as if something has gone wrong, and the data that's supposed to
be used with the tutorial is no longer shipped with 2.31. Or else I'm
doing something wrong in the installation process. Could you advise?
Thanks for your help,
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Pkg-ExpPsy-PyMVPA