Why Is My Classifier Discriminatory?

TL;DR

This paper argues that classifier unfairness often stems from data issues rather than model bias, and emphasizes improving data quality to reduce discrimination without sacrificing accuracy.

Contribution

It introduces a decomposition of discrimination metrics into bias, variance, and noise, advocating data collection as a primary method to mitigate unfairness.

Findings

Data collection can reduce discrimination without losing accuracy.

Decomposition of discrimination metrics helps identify sources of unfairness.

Case studies confirm the importance of data quality in fairness.

Abstract

Recent attempts to achieve fairness in predictive models focus on the balance between fairness and accuracy. In sensitive applications such as healthcare or criminal justice, this trade-off is often undesirable as any increase in prediction error could have devastating consequences. In this work, we argue that the fairness of predictions should be evaluated in context of the data, and that unfairness induced by inadequate samples sizes or unmeasured predictive variables should be addressed through data collection, rather than by constraining the model. We decompose cost-based metrics of discrimination into bias, variance, and noise, and propose actions aimed at estimating and reducing each term. Finally, we perform case-studies on prediction of income, mortality, and review ratings, confirming the value of this analysis. We find that data collection is often a means to reduce discrimination without sacrificing accuracy.