Fair Data Adaptation with Quantile Preservation

TL;DR

This paper introduces a practical data adaptation method using quantile preservation within causal models to promote fairness in classification and regression, ensuring fairness criteria are met even under model misspecification.

Contribution

The paper proposes a novel data adaptation technique based on quantile preservation in causal models, which guarantees certain fairness notions despite counterfactual model uncertainties.

Findings

Method guarantees population fairness notions under misspecification.

Implementation based on Random Forests demonstrates practical effectiveness.

Applicable to simulated and real-world datasets.

Abstract

Fairness of classification and regression has received much attention recently and various, partially non-compatible, criteria have been proposed. The fairness criteria can be enforced for a given classifier or, alternatively, the data can be adapated to ensure that every classifier trained on the data will adhere to desired fairness criteria. We present a practical data adaption method based on quantile preservation in causal structural equation models. The data adaptation is based on a presumed counterfactual model for the data. While the counterfactual model itself cannot be verified experimentally, we show that certain population notions of fairness are still guaranteed even if the counterfactual model is misspecified. The precise nature of the fulfilled non-causal fairness notion (such as demographic parity, separation or sufficiency) depends on the structure of the underlying causal model and the choice of resolving variables. We describe an implementation of the proposed data adaptation procedure based on Random Forests and demonstrate its practical use on simulated and real-world data.