mne_rsa.rsa_array#

mne_rsa.rsa_array(X, rdm_model, patches=None, data_rdm_metric='correlation', data_rdm_params={}, rsa_metric='spearman', ignore_nan=False, y=None, labels_X=None, labels_rdm_model=None, n_folds=1, n_jobs=1, verbose=False)[source]#

Perform RSA on an array of data, possibly in a searchlight pattern.

Parameters:

Xndarray, shape (n_items, n_series, n_times)

An array containing the data.

rdm_modelndarray, shape (n, n) | (n * (n - 1) // 2,) | list of ndarray

The model RDM, see compute_rdm(). For efficiency, you can give it in condensed form, meaning only the upper triangle of the matrix as a vector. See scipy.spatial.distance.squareform(). To perform RSA against multiple models at the same time, supply a list of model RDMs.

Use compute_rdm() to compute RDMs.

patchesgenerator of tuples | None

Searchlight patches as generated by searchlight. If None, no searchlight is used. Defaults to None.

data_rdm_metricstr

The metric to use to compute the data RDMs. This can be any metric supported by the scipy.distance.pdist function. Defaults to ‘correlation’.

data_rdm_paramsdict

Extra arguments for the distance metric used to compute the RDMs. Refer to scipy.spatial.distance for a list of all other metrics and their arguments. Defaults to an empty dictionary.

rsa_metricstr

The RSA metric to use to compare the RDMs. Valid options are:

‘spearman’ for Spearman’s correlation (the default)
‘pearson’ for Pearson’s correlation
‘kendall-tau-a’ for Kendall’s Tau (alpha variant)
‘partial’ for partial Pearson correlations
‘partial-spearman’ for partial Spearman correlations
‘regression’ for linear regression weights

Defaults to ‘spearman’.

ignore_nanbool

Whether to treat NaN’s as missing values and ignore them when computing the distance metric. Defaults to False.

Added in version 0.8.

yndarray of int, shape (n_items,) | None

Deprecated, use labels_X and labels_rdm_model instead. For each item, a number indicating the class to which the item belongs. Defaults to None, in which case labels_X is used.

labels_Xlist | None

For each element in X (=the first dimension), a label that identifies the item to which it corresponds. This is used in combination with labels_rdm_model to align the data and model RDMs before comparing them. Multiple elements in X may correspond to the same item, in which case they should have the same label and will either be averaged when computing the data RDM (n_folds=1) or used for cross-validation (n_folds>1). Labels may be of any python type that can be compared with == (int, float, string, tuple, etc). By default (None), the integers 0:len(X) are used as labels.

Added in version 0.10.

labels_rdm_model: list | None

For each row in rdm_model, a label that identifies the item to which it corresponds. This is used in combination with labels_X to align the data and model RDMs before comparing them. Each row should have a unique label. Labels may be of any python type that can be compared with == (int, float, string, tuple, etc). By default (None), the integers 0:n_rows are used as labels.

Added in version 0.10.

n_foldsint | sklearn.model_selection.BaseCrollValidator | None

Number of cross-validation folds to use when computing the distance metric. Folds are created based on the y parameter. Specify None to use the maximum number of folds possible, given the data. Alternatively, you can pass a Scikit-Learn cross validator object (e.g. sklearn.model_selection.KFold) to assert fine-grained control over how folds are created. Defaults to 1 (no cross-validation).

n_jobsint

The number of processes (=number of CPU cores) to use. Specify -1 to use all available cores. Defaults to 1.

verbosebool

Whether to display a progress bar. In order for this to work, you need the tqdm python module installed. Defaults to False.

Returns:

rsa_valsndarray, shape ([n_model_rdms,] [n_series,] [n_times,]): The RSA value for each searchlight patch. When spatial_radius is set to None, there will only be no n_series dimension. When temporal_radius is set to None, there will be no time dimension. When multiple models have been supplied, the first dimension will contain RSA results for each model.