A Consequentialist Critique of Binary Classification Evaluation Practices
This work addresses a foundational problem in ML evaluation for researchers and practitioners, offering incremental improvements by mapping metrics to use cases and providing tools to bridge the gap between theory and practice.
The paper tackles the misalignment between binary classification evaluation practices and consequentialist decision-making, revealing a strong preference for top-K metrics or fixed thresholds in major conferences despite theoretical support for proper scoring rules like Brier scores, and it introduces a Python package to promote their adoption.
ML-supported decisions, such as ordering tests or determining preventive custody, often involve binary classification based on probabilistic forecasts. Evaluation frameworks for such forecasts typically consider whether to prioritize independent-decision metrics (e.g., Accuracy) or top-K metrics (e.g., Precision@K), and whether to focus on fixed thresholds or threshold-agnostic measures like AUC-ROC. We highlight that a consequentialist perspective, long advocated by decision theorists, should naturally favor evaluations that support independent decisions using a mixture of thresholds given their prevalence, such as Brier scores and Log loss. However, our empirical analysis reveals a strong preference for top-K metrics or fixed thresholds in evaluations at major conferences like ICML, FAccT, and CHIL. To address this gap, we use this decision-theoretic framework to map evaluation metrics to their optimal use cases, along with a Python package, briertools, to promote the broader adoption of Brier scores. In doing so, we also uncover new theoretical connections, including a reconciliation between the Brier Score and Decision Curve Analysis, which clarifies and responds to a longstanding critique by (Assel, et al. 2017) regarding the clinical utility of proper scoring rules.