[pymvpa] pymvpa help

[Yaroslav Halchenko]

On Sun, 14 Feb 2016, ibinliu at sina.com wrote:

>    My experiment contained 2 tasks (only 1 run). 

it is quite unfortunate that you have only a single run.  but let's
see...

> I performed single trial
>    analysis to generate a beta map for every trial. I'd like to run
>    classification analysis on these beta maps.

how many tries for each of those 2 conditions do you have?

how many subjects do you have?

what was design (block or even-related, ISI, randomized how)?

>    The first column of attributes was taskA, taskA, ..., taskB, taskB... and
>    the second column were 0s (since I only got 1 run).
>    Is this right?
>    For each suject, I merged all the beta images into a single 4D file and
>    used attr = SampleAttributes('attributes.txt') and
>    ds=fmri_dataset(samples=a**bold.nii.gza**, targets=attr.targets,
>    chunks=attr.chunks, mask=a**mask.nii.gza**) to load data.

is that bold.nii.gz already those beta maps or original data?

>    How should I perform normalization? just zscore(ds)?

should do it

>    I want to run cross-validation (say 5 fold or leave-one-out), how could I
>    set the partitioner?

It is somewhat depends on above questions
-- 
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