mne.stats.permutation_cluster_test(X, threshold=None, n_permutations=1024, tail=0, stat_fun=<function f_oneway>, connectivity=None, verbose=None, n_jobs=1, seed=None, max_step=1, exclude=None, step_down_p=0, t_power=1, out_type=’mask’, check_disjoint=False, buffer_size=1000)[source]

Cluster-level statistical permutation test.

For a list of nd-arrays of data, e.g. 2d for time series or 3d for time-frequency power values, calculate some statistics corrected for multiple comparisons using permutations and cluster level correction. Each element of the list X contains the data for one group of observations. Randomized data are generated with random partitions of the data.


X : list

List of nd-arrays containing the data. Each element of X contains the samples for one group. First dimension of each element is the number of samples/observations in this group. The other dimensions are for the size of the observations. For example if X = [X1, X2] with X1.shape = (20, 50, 4) and X2.shape = (17, 50, 4) one has 2 groups with respectively 20 and 17 observations in each. Each data point is of shape (50, 4).

threshold : float | dict | None

If threshold is None, it will choose a t-threshold equivalent to p < 0.05 for the given number of (within-subject) observations. If a dict is used, then threshold-free cluster enhancement (TFCE) will be used.

n_permutations : int

The number of permutations to compute.

tail : -1 or 0 or 1 (default = 0)

If tail is 1, the statistic is thresholded above threshold. If tail is -1, the statistic is thresholded below threshold. If tail is 0, the statistic is thresholded on both sides of the distribution.

stat_fun : callable

function called to calculate statistics, must accept 1d-arrays as arguments (default: scipy.stats.f_oneway).

connectivity : sparse matrix.

Defines connectivity between features. The matrix is assumed to be symmetric and only the upper triangular half is used. Default is None, i.e, a regular lattice connectivity.

verbose : bool, str, int, or None

If not None, override default verbose level (see mne.verbose() and Logging documentation for more).

n_jobs : int

Number of permutations to run in parallel (requires joblib package).

seed : int or None

Seed the random number generator for results reproducibility.

max_step : int

When connectivity is a n_vertices x n_vertices matrix, specify the maximum number of steps between vertices along the second dimension (typically time) to be considered connected. This is not used for full or None connectivity matrices.

exclude : boolean array or None

Mask to apply to the data to exclude certain points from clustering (e.g., medial wall vertices). Should be the same shape as X. If None, no points are excluded.

step_down_p : float

To perform a step-down-in-jumps test, pass a p-value for clusters to exclude from each successive iteration. Default is zero, perform no step-down test (since no clusters will be smaller than this value). Setting this to a reasonable value, e.g. 0.05, can increase sensitivity but costs computation time.

t_power : float

Power to raise the statistical values (usually f-values) by before summing (sign will be retained). Note that t_power == 0 will give a count of nodes in each cluster, t_power == 1 will weight each node by its statistical score.

out_type : str

For arrays with connectivity, this sets the output format for clusters. If ‘mask’, it will pass back a list of boolean mask arrays. If ‘indices’, it will pass back a list of lists, where each list is the set of vertices in a given cluster. Note that the latter may use far less memory for large datasets.

check_disjoint : bool

If True, the connectivity matrix (or list) will be examined to determine of it can be separated into disjoint sets. In some cases (usually with connectivity as a list and many “time” points), this can lead to faster clustering, but results should be identical.

buffer_size: int or None

The statistics will be computed for blocks of variables of size “buffer_size” at a time. This is option significantly reduces the memory requirements when n_jobs > 1 and memory sharing between processes is enabled (see set_cache_dir()), as X will be shared between processes and each process only needs to allocate space for a small block of variables.


T_obs : array of shape [n_tests]

T-statistic observed for all variables.

clusters : list

List type defined by out_type above.

cluster_pv : array

P-value for each cluster

H0 : array of shape [n_permutations]

Max cluster level stats observed under permutation.


Reference: Cluster permutation algorithm as described in Maris/Oostenveld (2007), “Nonparametric statistical testing of EEG- and MEG-data” Journal of Neuroscience Methods, Vol. 164, No. 1., pp. 177-190. doi:10.1016/j.jneumeth.2007.03.024