Fair Decision Rules for Binary Classification

TL;DR

This paper introduces a method for creating fair, interpretable Boolean rule sets for binary classification that satisfy fairness constraints like equality of opportunity and equalized odds, using an efficient column generation approach.

Contribution

It formulates the fair rule set learning as an integer program with a novel column generation framework, enabling the handling of large datasets and stricter fairness constraints.

Findings

Achieves fairer rule sets with modest accuracy trade-offs.

Handles large datasets efficiently with heuristics.

Outperforms existing fair classifiers in fairness constraints.

Abstract

In recent years, machine learning has begun automating decision making in fields as varied as college admissions, credit lending, and criminal sentencing. The socially sensitive nature of some of these applications together with increasing regulatory constraints has necessitated the need for algorithms that are both fair and interpretable. In this paper we consider the problem of building Boolean rule sets in disjunctive normal form (DNF), an interpretable model for binary classification, subject to fairness constraints. We formulate the problem as an integer program that maximizes classification accuracy with explicit constraints on two different measures of classification parity: equality of opportunity and equalized odds. Column generation framework, with a novel formulation, is used to efficiently search over exponentially many possible rules. When combined with faster heuristics, our method can deal with large data-sets. Compared to other fair and interpretable classifiers, our method is able to find rule sets that meet stricter notions of fairness with a modest trade-off in accuracy.