directed_hausdorff_distance¶

kooplearn.metrics.directed_hausdorff_distance(pred: ndarray, reference: ndarray)[source]¶

One-sided directed Hausdorff distance between two 1D sets. Useful for computing distances between estimated eigenvalues

Calculates the directed Hausdorff distance \(\vec{H}(A, B) = \max_{a \in A} \min_{b \in B} \|a - b\|_p\) where \(A\) is the set of points in pred and \(B\) is the set of points in reference. The current implementation uses the \(L_1\) norm: \(\|a - b\|_1 = |a - b|\).

Parameters:

pred (numpy.ndarray) – The set of predicted points \(A\). Must be a 1D array.
reference (numpy.ndarray) – The set of reference points \(B\). Must be a 1D array.

Returns:

The directed Hausdorff distance between pred``and ``reference.

Return type:

float

Raises:

AssertionError – If pred or reference are not 1-dimensional arrays.

Examples

import numpy as np
from kooplearn.metrics import directed_hausdorff_distance
pred = np.array([1, 5, 6])
reference = np.array([2, 4, 7])
directed_hausdorff_distance(pred, reference)
# Will print np.float64(1.0)