moabb.analysis.meta_analysis.find_significant_differences

moabb.analysis.meta_analysis.find_significant_differences(df, perm_cutoff=20)

Compute differences between pipelines across datasets.

Compute matrices of p-values and effects for all algorithms over all datasets via combined p-values and combined effects methods

Parameters

• df (DataFrame) – Table of effect and p-values for each dataset and all pipelines, returned by compute_dataset_statistics

• perm_cutoff (int, default=20) – threshold value to stop using permutation tests, which can be very expensive computationally, using Wilcoxon rank-sum test instead

Returns

• dfP (DataFrame of shape (n_pipelines, n_pipelines)) – p-values per algorithm pairs

• dfT (DataFrame of shape (n_pipelines, n_pipelines)) – signed standardized mean differences