Is your model predicting the past?

TL;DR

This paper introduces a framework with statistical tests called backward baselines to distinguish whether machine learning models predict future outcomes or simply recite past patterns, supported by theory and empirical evaluation.

Contribution

It proposes a novel set of statistical tests for auditing models to determine if they predict future or just reflect past data, with practical guidance and empirical validation.

Findings

Backward baselines effectively distinguish prediction types.

Theoretical guidance aids interpretation of baseline results.

Empirical tests on survey data demonstrate practical utility.

Abstract

When does a machine learning model predict the future of individuals and when does it recite patterns that predate the individuals? In this work, we propose a distinction between these two pathways of prediction, supported by theoretical, empirical, and normative arguments. At the center of our proposal is a family of simple and efficient statistical tests, called backward baselines, that demonstrate if, and to what extent, a model recounts the past. Our statistical theory provides guidance for interpreting backward baselines, establishing equivalences between different baselines and familiar statistical concepts. Concretely, we derive a meaningful backward baseline for auditing a prediction system as a black box, given only background variables and the system's predictions. Empirically, we evaluate the framework on different prediction tasks derived from longitudinal panel surveys, demonstrating the ease and effectiveness of incorporating backward baselines into the practice of machine learning.