pitcp.utils.coverage_gap

coverage_gap(labels, covered)

Computes the range of empirical coverage across labeled groups.

Parameters:

• labels (np.typing.ArrayLike) – Group labels with shape (n_samples,).

• covered (np.typing.ArrayLike) – Boolean mask with shape (n_samples,) or (n_samples, n_levels).

Returns:

Maximum minus minimum group coverage. A scalar is returned

for one coverage mask and an array for multiple masks.

Return type:

float | np.ndarray

Raises:

ValueError – If inputs are empty, have invalid dimensions, or contain different sample counts.