Skip to content

fcma: compute_correlation returns zeros for NaNs #400

@snastase

Description

@snastase

I just noticed that brainiak.fcma.util.compute_correlation returns float zeros when computing the correlation between vectors containing NaNs, which is inconsistent with other correlation computations in Python like np.corrcoef and scipy.stats.pearsonr (which return np.nan). Is this the desired output? Or is it necessary for the high-speed BLAS implementation? Might not be worth bothering with—was just curious.

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