Causal Strategic Classification: A Tale of Two Shifts

TL;DR

This paper introduces a causal strategic classification framework that accounts for changing true outcomes due to user behavior, addressing distribution shifts and proposing a novel learning algorithm validated on synthetic data.

Contribution

It removes the assumption that features do not affect outcomes, modeling causal effects and developing an algorithm to handle two types of distribution shifts in strategic classification.

Findings

The proposed method effectively balances distribution shifts caused by strategic behavior.

Experiments show improved robustness and accuracy over traditional models.

The approach is validated on synthetic and semi-synthetic datasets.

Abstract

When users can benefit from certain predictive outcomes, they may be prone to act to achieve those outcome, e.g., by strategically modifying their features. The goal in strategic classification is therefore to train predictive models that are robust to such behavior. However, the conventional framework assumes that changing features does not change actual outcomes, which depicts users as "gaming" the system. Here we remove this assumption, and study learning in a causal strategic setting where true outcomes do change. Focusing on accuracy as our primary objective, we show how strategic behavior and causal effects underlie two complementing forms of distribution shift. We characterize these shifts, and propose a learning algorithm that balances between these two forces and over time, and permits end-to-end training. Experiments on synthetic and semi-synthetic data demonstrate the utility of our approach.