hausdorff_metric

hausdorff_metric(changepoints_true: ArrayLike, changepoints_pred: ArrayLike, *, max_distance: float | None = None) → float

Compute the Hausdorff distance between two changepoint sets.

Measures the maximum distance from any true changepoint to its nearest predicted changepoint, and vice versa. Lower is better.

Parameters:

changepoints_truearray-like of shape (n_changepoints_true,)

True changepoint indices, as returned by predict_changepoints().

changepoints_predarray-like of shape (n_changepoints_pred,)

Predicted changepoint indices, as returned by predict_changepoints().

max_distancefloat | None, default=None

Cap on the returned distance. If None, no cap is applied.

Returns:

float

Hausdorff distance. Returns 0.0 if both have no changepoints, inf (or max_distance) if exactly one has no changepoints.

Examples

>>> hausdorff_metric([10, 20], [12, 20])
2.0