nispace.stats.effectsize.cohen_paired_nan

nispace.stats.effectsize.cohen_paired_nan(a, b)[source]

NaN-aware Cohen’s d for paired/dependent samples (see cohen_paired).

Parameters:

  • a (array_like, shape (n_obs, n_features)) – Same convention as cohen_paired; must have identical shape. NaN pairs are excluded per-column via nanmean/nanstd on a - b.

  • b (array_like, shape (n_obs, n_features)) – Same convention as cohen_paired; must have identical shape. NaN pairs are excluded per-column via nanmean/nanstd on a - b.

Returns:

d

Return type:

np.ndarray, shape (n_features,)

Raises:

ValueError – If a.shape != b.shape.

Notes

Not called anywhere in NiSpace’s internal pipeline: Y_transform= “pairedcohen(a,b)” dispatches to the numba-jitted cohen_paired_nan_fast (core/transform_y.py), not this function.

References

[39].