rxmc.model_comparison.coverage_curve#

rxmc.model_comparison.coverage_curve(draws, y, levels=None) → ndarray[source]#

Empirical coverage of central predictive intervals at each nominal level.

Parameters:

  • draws (array_like, shape (n_draws, n_pts))

  • y (array_like, shape (n_pts,)) – The data the draws are checked against (same space as draws).

  • levels (array_like, optional) – Nominal central-interval probabilities in (0, 1). Defaults to 49 levels from 0.02 to 0.98 (_DEFAULT_LEVELS).

Returns:

np.ndarray – Fraction of points inside the central level interval, per level.