EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

1655. Fair treatment allocation via tree ensembles

Invited abstract in session MB-27: On Mathematical Optimization for Explainable and Fair Machine Learning, stream Mathematical Optimization for XAI.

Monday, 10:30-12:00
Room: 047 (building: 208)

Authors (first author is the speaker)

1. Kseniia Kurishchenko
Economics, Copenhagen Business School

Abstract

In the treatment allocation problem, a decision maker needs to decide who will receive the treatment. To define the individuals that benefit the most, a machine learning procedure is often used to predict the treatment effect. With these predictions, a model is built to make the allocation decisions. Unfortunately, the data used to build such a model may be discriminating against a group defined by a sensitive attribute such as gender or age. If not carefully trained, the model may provide unfair results, unequally allocating treatment to individuals in the sensitive and non-sensitive groups.
In this presentation, I introduce an unfairness measure that can be applied to treatment allocation problems. I propose to measure unfairness as the difference between the average treatment effects in the sensitive group and the non-sensitive group. I introduce a Mathematical Optimization model to have accurate treatment effect predictions and a good level of fairness, which will be the basis for the treatment allocation in forthcoming individuals.

Keywords

Status: accepted

Back to the list of papers