Can a calibration metric be both testable and actionable?

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Rossellini, Raphael, Soloff, Jake A., Barber, Rina Foygel, Ren, Zhimei, Willett, Rebecca
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908478200610816
author Rossellini, Raphael
Soloff, Jake A.
Barber, Rina Foygel
Ren, Zhimei
Willett, Rebecca
author_facet Rossellini, Raphael
Soloff, Jake A.
Barber, Rina Foygel
Ren, Zhimei
Willett, Rebecca
contents Forecast probabilities often serve as critical inputs for binary decision making. In such settings, calibration$\unicode{x2014}$ensuring forecasted probabilities match empirical frequencies$\unicode{x2014}$is essential. Although the common notion of Expected Calibration Error (ECE) provides actionable insights for decision making, it is not testable: it cannot be empirically estimated in many practical cases. Conversely, the recently proposed Distance from Calibration (dCE) is testable, but it is not actionable since it lacks decision-theoretic guarantees needed for high-stakes applications. To resolve this question, we consider Cutoff Calibration Error, a calibration measure that bridges this gap by assessing calibration over intervals of forecasted probabilities. We show that Cutoff Calibration Error is both testable and actionable, and we examine its implications for popular post-hoc calibration methods, such as isotonic regression and Platt scaling.
format Preprint
id arxiv_https___arxiv_org_abs_2502_19851
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Can a calibration metric be both testable and actionable?
Rossellini, Raphael
Soloff, Jake A.
Barber, Rina Foygel
Ren, Zhimei
Willett, Rebecca
Methodology
Machine Learning
Forecast probabilities often serve as critical inputs for binary decision making. In such settings, calibration$\unicode{x2014}$ensuring forecasted probabilities match empirical frequencies$\unicode{x2014}$is essential. Although the common notion of Expected Calibration Error (ECE) provides actionable insights for decision making, it is not testable: it cannot be empirically estimated in many practical cases. Conversely, the recently proposed Distance from Calibration (dCE) is testable, but it is not actionable since it lacks decision-theoretic guarantees needed for high-stakes applications. To resolve this question, we consider Cutoff Calibration Error, a calibration measure that bridges this gap by assessing calibration over intervals of forecasted probabilities. We show that Cutoff Calibration Error is both testable and actionable, and we examine its implications for popular post-hoc calibration methods, such as isotonic regression and Platt scaling.
title Can a calibration metric be both testable and actionable?
topic Methodology
Machine Learning
url https://arxiv.org/abs/2502.19851