Skip to content

make it easier to specify adjacency in permutation_cluster_test #9364

@drammock

Description

@drammock

I tried to do a time-frequency permutation_cluster_test over 126 channels for different observations and for two groups.
So I have 4D data (observations, times, frequencies, channels) for two groups of users but it didn't work.

adjacency = sparse.csr_matrix(np.zeros((126,126)))
sample = np.zeros((4,100,6,126))

print(sample.shape)
print(adjacency.shape)

F_obs, clusters, cluster_pv, H0 = mne.stats.permutation_cluster_test(X=[sample, sample], adjacency=adjacency)

the error raised is

ValueError: adjacency (len 126) must be of the correct size, i.e. be equal to or evenly divide the number of tests (75600). If adjacency was computed for a source space, try using the fwd["src"] or inv["src"] as some original source space vertices can be excluded during forward computation

at line 606 of "mne/stats/cluster_level.py" file.
(and 75600 is evenly dividable by 126, but "got_times" is different from "n_times" for the dimension more)

It seems like at most there can be three dimensions and not four.
I thought that with this version it was possible to do it.

Is it not?

Best,
Giovanni

Originally posted by @GiovanniGr in #7915 (comment)

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions