bpca._utils.compute_variance_explained

bpca._utils.compute_variance_explained(X, usage, loadings)

Compute variance explained by each component.

Uses leave-one-out contributions normalized to sum to total R². This handles non-orthogonal components correctly.

Parameters:

• X (ndarray) – Original data matrix (n_obs, n_var)

• usage (ndarray) – Score matrix (n_obs, n_components)

• loadings (ndarray) – Loading matrix (n_components, n_var)

Return type:

ndarray

Returns:

tuple[np.ndarray, np.ndarray] Variance explained per component (sorted descending) and sort order