[Mne_analysis] question about permutation_cluster_test

[Dan McCloy]

        External Email - Use Caution        

Hi Fleur,
The **last dimension** of the arrays in X should be the dimension represented by `connectivity`.  So this is why it must be (n_subjs x n_freq x n_channels), instead of (n_subjs x n_channels x n_freq).  I agree that this is not explained well in [the docstring](https://mne.tools/dev/generated/mne.stats.permutation_cluster_test.html). Would you like to make a pull request suggesting a docstring improvement?

-- dan
Daniel McCloy
https://dan.mccloy.info
Research Scientist
Institute for Learning and Brain Sciences
University of Washington

‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
On Wednesday, April 15, 2020 7:28 AM, Fleur GAUDFERNAU <fleur.gaudfernau at pasteur.fr> wrote:

>         External Email - Use Caution
>
> Hello,
>
> I juste have a quick question regarding mne.stats.permutation_cluster_test.
>
> When I give it two arrays X1 and X2 of shape (n subjects x n channels x n freq), I have the following error: ValueError: connectivity (len 20484) must be of the correct size, i.e. be equal to or evenly divide the number of tests (2580984).
>
> The error disappear if I reshape X1 and X2 into shape (n subjects x n freq x nchannels).
>
> I was wondering why because I did not see anything about that in the documentation of permutation_cluster_test.
>
> Thanks in advance!
>
> Kind regards,
>
> Fleur
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20200416/910894b4/attachment.html