holisticai.bias.metrics.social_fairness_ratio

holisticai.bias.metrics.social_fairness_ratio(group_a, group_b, data, centroids)

Social Fairness Ratio

Given a centroid based clustering, this function compute the average distance to the nearest centroid for both groups. The metric is the ratio of the resulting distance for group_a to group_b.

Interpretation

A value of 1 is desired. Lower values indicate the group_a is on average closer to the respective centroids. Higher values indicate that group_a is on average further from the respective centroids.

Parameters

group_aarray-like

Group membership vector (binary)

group_barray-like

Group membership vector (binary)

datamatrix-like

Data matrix of shape (num_inst, dim)

centroidsmatrix-like

Centroids (centers) of shape (num_centroids, dim)

Returns

float

Social Fairness Ratio

Examples

>>> import numpy as np
>>> from holisticai.bias.metrics import social_fairness_ratio
>>> group_a = np.array([1, 1, 1, 1, 0, 0, 0, 0, 0, 0])
>>> group_b = np.array([0, 0, 0, 0, 1, 1, 1, 1, 1, 1])
>>> data = np.array(
...     [
...         [-1, 1],
...         [1, 1],
...         [1, 1],
...         [0, -1],
...         [-1, 1],
...         [-1, 1],
...         [-1, 1],
...         [-1, 1],
...         [1, 1],
...         [0, -1],
...     ]
... )
>>> centroids = np.array([[-2, 1], [1, 2], [0, -2]])
>>> social_fairness_ratio(group_a, group_b, data, centroids)
1.0