The Need for Transparent Demographic Group Trade-Offs in Credit Risk and Income Classification

Abstract

Prevalent methodology towards constructing fair machine learning (ML) systems, is to enforce a strict equality metric for demographic groups based on protected attributes like race and gender. While definitions of fairness in philosophy are varied, mitigating bias in ML classifiers often relies on demographic parity-based constraints across sub-populations. However, enforcing such constraints blindly can lead to undesirable trade-offs between group-level accuracy if groups possess different underlying sampled population metrics, an occurrence that is surprisingly common in real-world applications like credit risk and income classification. Similarly, attempts to relax hard constraints may lead to unintentional degradation in classification performance, without benefit to any demographic group. In these increasingly likely scenarios, we make the case for transparent human intervention in making the trade-offs between the accuracies of demographic groups. We propose that transparency in trade-offs between demographic groups should be a key tenet of ML design and implementation. Our evaluation demonstrates that a transparent human-in-the-loop trade-off technique based on the Pareto principle increases both overall and group-level accuracy by 9.5% and 9.6% respectively, in two commonly explored UCI datasets for credit risk and income classification.