fairlearn.reductions.ErrorRateParity

class fairlearn.reductions.ErrorRateParity(*, difference_bound=None, ratio_bound=None, ratio_bound_slack=0.0)

Implementation of error rate parity as a moment.

A classifier \(h(X)\) satisfies error rate parity if

\[P[h(X) \ne Y | A = a] = P[h(X) \ne Y] \; \forall a\]

This implementation of UtilityParity defines a single event, all. Consequently, the prob_event pandas.Series will only have a single entry, which will be equal to 1.

The index property will have twice as many entries (corresponding to the Lagrange multipliers for positive and negative constraints) as there are unique values for the sensitive feature.

The UtilityParity.signed_weights() method will compute the costs according to Example 3 of Agarwal et al.[1]. However, in this scenario, g = abs(h(x)-y), rather than g = h(x)

This Moment also supports control features, which can be used to stratify the data, with the constraint applied within each stratum, but not between strata.

Read more in the User Guide.

bound()

Return bound vector.

Returns:

pandas.Series

a vector of bound values corresponding to all constraints

default_objective()

Return the default objective for moments of this kind.

gamma(predictor)

Calculate the degree to which constraints are currently violated by the predictor.

load_data(X, y, *, sensitive_features, control_features=None)

Load the specified data into the object.

project_lambda(lambda_vec)

Return the projected lambda values.

i.e., returns lambda which is guaranteed to lead to the same or higher value of the Lagrangian compared with lambda_vec for all possible choices of the classifier, h.

signed_weights(lambda_vec)

Compute the signed weights.

Uses the equations for \(C_i^0\) and \(C_i^1\) as defined in Section 3.2 of Agarwal et al.[1] in the ‘best response of the Q-player’ subsection to compute the signed weights to be applied to the data by the next call to the underlying estimator.

Parameters:

lambda_vecpandas.Series

The vector of Lagrange multipliers indexed by index

short_name = 'ErrorRateParity'

property total_samples

Return the number of samples in the data.