[pymvpa] Hyperalignment error: "For now do not handle invariant in time datasets"

[Yaroslav O Halchenko]

On Sun, 04 Nov 2018, Chang Wenshuo wrote:

>  Dear all,

>  I am running hyperalignment between-subject analysis. Before calling Hyperalignment(), I stacked the

>  datasets across the subjects with all runs and did Anova-based feature selection. Hyperalignment() fails

>  for the dataset with the error message: "For now do not handle invariant in time datasets".



>  It would be great if somebody has any kind of information about what causes this issue, and ways to
>  solve it. The problem is not fixed after I removed invariant features by using remove_invariant_features()
>  or set the svd in ProcrusteanMapper() to 'dgesvd'.

remove_invariant_features works across runs, and probably what happens
in your code that only some runs still somehow have somehow invariant
voxels.    are you masking with different masks in each run?


>  My codes are as follow:

>  # split in training and testing sets
>  ds_train = dscon[np.array([r != test_run for r in dscon.sa.chunks], dtype='bool'),:]
>  ds_test = dscon[np.array([r == test_run for r in dscon.sa.chunks], dtype='bool'),:]
>  ds_train = remove_invariant_features(ds_train)
-- 
Yaroslav O. Halchenko
Center for Open Neuroscience     http://centerforopenneuroscience.org
Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755
Phone: +1 (603) 646-9834                       Fax: +1 (603) 646-1419
WWW:   http://www.linkedin.com/in/yarik