Where Is a Prediction Model Miscalibrated, and for Whom? Local Audits in Outcome-Learned Pockets
Abstract
Overall and group-level averages, and ranking metrics such as the AUC, can hide sets of observations for which a prediction model systematically over- or under-predicts. We present an audit that asks where a model is miscalibrated (in which sets of similar observations it over- or under-predicts on average) and for whom (whether its average prediction error differs between demographic groups, such as by race or sex, within the same set). It groups observations into pockets using random-forest similarity, which weighs each feature by its importance for predicting the outcome, with a ridge-regression grouping alongside. Pockets are learned on one sample and tested on another, with simultaneous confidence intervals that allow for correlation within units such as hospitals; the audit needs only features, outcomes and fixed predictions. In 17 simulation scenarios with an error planted for one group, the combined audit located it better than the strongest alternative grouping in 15, with false-alarm rates near or below 5%; with a local error planted for everyone, it won in 12 of 15 scenarios at the smaller error size and lost when the error was unrelated to the outcome. In public synthetic data from the care-management algorithm studied by Obermeyer et al., it found no detectable overall error but under-prediction of illness in the sickest pockets and over-prediction elsewhere, and distinguished Black patients' concentration where the score under-predicted for everyone from a Black-White gap within pockets. For APACHE IVa, an exploratory analysis located its previously reported over-prediction of ICU mortality in a mostly ventilated pocket and a low-risk pocket, with none detected in a moderate-risk one. Two chest X-ray classifiers showed offsetting local errors behind no detectable overall error. Code: https://github.com/modenesi/pocket-audit
Related
- Gradient boosting machine for survival prediction in studies with general failure-time outcome-dependent sampling design.
- BRP-GABLE: An Interpretable Rule-Based Prediction Framework for Multiple Outcome Types
- Food-System Stress Testing: Prediction Tasks Under Alternative Outcome Definitions
- Self-intersection Points of Billiard Trajectories in a Square with Small Pockets
- Risk-Limiting Audits for Parliamentary Majorities
BibTeX
@misc{modenesi2026where,
title = {{Where Is a Prediction Model Miscalibrated, and for Whom? Local Audits in Outcome-Learned Pockets}},
author = {Bernardo Modenesi and Taoyi Chen and Sima Najafzadehkhoei and Lucia Wang and Ameya Diwan and Beatrice Knudsen},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.02639},
eprint = {2611.02639},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.02639},
}
RIS
TY - UNPB
TI - Where Is a Prediction Model Miscalibrated, and for Whom? Local Audits in Outcome-Learned Pockets
AU - Bernardo Modenesi
AU - Taoyi Chen
AU - Sima Najafzadehkhoei
AU - Lucia Wang
AU - Ameya Diwan
AU - Beatrice Knudsen
PY - 2026
DA - 2026/11/02
PB - arXiv
DO - 10.48550/arxiv.2611.02639
UR - https://arxiv.org/abs/2611.02639
AB - Overall and group-level averages, and ranking metrics such as the AUC, can hide sets of observations for which a prediction model systematically over- or under-predicts. We present an audit that asks where a model is miscalibrated (in which sets of similar observations it over- or under-predicts on average) and for whom (whether its average prediction error differs between demographic groups, such as by race or sex, within the same set). It groups observations into pockets using random-forest similarity, which weighs each feature by its importance for predicting the outcome, with a ridge-regression grouping alongside. Pockets are learned on one sample and tested on another, with simultaneous confidence intervals that allow for correlation within units such as hospitals; the audit needs only features, outcomes and fixed predictions. In 17 simulation scenarios with an error planted for one group, the combined audit located it better than the strongest alternative grouping in 15, with false-alarm rates near or below 5%; with a local error planted for everyone, it won in 12 of 15 scenarios at the smaller error size and lost when the error was unrelated to the outcome. In public synthetic data from the care-management algorithm studied by Obermeyer et al., it found no detectable overall error but under-prediction of illness in the sickest pockets and over-prediction elsewhere, and distinguished Black patients' concentration where the score under-predicted for everyone from a Black-White gap within pockets. For APACHE IVa, an exploratory analysis located its previously reported over-prediction of ICU mortality in a mostly ventilated pocket and a low-risk pocket, with none detected in a moderate-risk one. Two chest X-ray classifiers showed offsetting local errors behind no detectable overall error. Code: https://github.com/modenesi/pocket-audit
ER -