A hierarchical decomposition for explaining ML performance discrepancies

TL;DR

This paper introduces a nonparametric hierarchical framework that provides detailed explanations for why machine learning performance varies across domains, enabling targeted interventions without needing causal knowledge.

Contribution

It develops a novel hierarchical decomposition method that offers both aggregate and variable-level insights into performance gaps, without relying on causal assumptions.

Findings

Provides debiased, efficient estimators for the decompositions

Enables statistical inference with valid confidence intervals

Improves understanding of domain-specific performance differences

Abstract

Machine learning (ML) algorithms can often differ in performance across domains. Understanding $\textit{why}$ their performance differs is crucial for determining what types of interventions (e.g., algorithmic or operational) are most effective at closing the performance gaps. Existing methods focus on $\textit{aggregate decompositions}$ of the total performance gap into the impact of a shift in the distribution of features $p(X)$ versus the impact of a shift in the conditional distribution of the outcome $p(Y|X)$; however, such coarse explanations offer only a few options for how one can close the performance gap. $\textit{Detailed variable-level decompositions}$ that quantify the importance of each variable to each term in the aggregate decomposition can provide a much deeper understanding and suggest much more targeted interventions. However, existing methods assume knowledge of the full causal graph or make strong parametric assumptions. We introduce a nonparametric hierarchical framework that provides both aggregate and detailed decompositions for explaining why the performance of an ML algorithm differs across domains, without requiring causal knowledge. We derive debiased, computationally-efficient estimators, and statistical inference procedures for asymptotically valid confidence intervals.