PreprintarXiv licencearXiv
Local Discrimination of Binary Classifiers via Localized AUC Functional
Abstract
Risk prediction models are frequently evaluated with AUC, which means that during evaluation we are blind to performance heterogeneity, which could lead to bad decision-making in clinics. In this work, we aim to change that by identifying patient cohorts for whom the risk prediction model might not be an accurate predictor. To separate these cohorts, we use a version of the weighted AUC, which we call Localized AUC or LAUC. Using this metric, we build a method that we coin LOCATOR. This method is able to identify cohorts of patients for whom the risk prediction model is inaccurate. We test this method both in real-world data and in simulations. We also build a statistical background for the use of Localized AUC for our purposes.
Related
- Rényi divergences and binary state discrimination error exponents for fermionic quasi-free states
- Bargmann Invariants Are Insufficient for Complete Local-Unitary Orbit Discrimination
- Advancing specific local wine identification through integration of ICP-MS mineral wine profiles and Machine Learning: A case study on discrimination among the ten Beaujolais crus
- Advancing Specific Local Wine Identification through Integration of ICP-MS Mineral Wine Profiles and Machine Learning: A Case Study on Discrimination Among the Ten Beaujolais Crus
- Sparse Moore-local realisations of binary irreducible polynomials on near-square lattice regions
BibTeX
@misc{ren2026local,
title = {{Local Discrimination of Binary Classifiers via Localized AUC Functional}},
author = {Wenye Ren and Anna Győrffy-Kerekes and Tancredi Dorsi and Olga Demler},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.31216},
eprint = {2610.31216},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.31216},
}
RIS
TY - UNPB
TI - Local Discrimination of Binary Classifiers via Localized AUC Functional
AU - Wenye Ren
AU - Anna Győrffy-Kerekes
AU - Tancredi Dorsi
AU - Olga Demler
PY - 2026
DA - 2026/10/22
PB - arXiv
DO - 10.48550/arxiv.2610.31216
UR - https://arxiv.org/abs/2610.31216
AB - Risk prediction models are frequently evaluated with AUC, which means that during evaluation we are blind to performance heterogeneity, which could lead to bad decision-making in clinics. In this work, we aim to change that by identifying patient cohorts for whom the risk prediction model might not be an accurate predictor. To separate these cohorts, we use a version of the weighted AUC, which we call Localized AUC or LAUC. Using this metric, we build a method that we coin LOCATOR. This method is able to identify cohorts of patients for whom the risk prediction model is inaccurate. We test this method both in real-world data and in simulations. We also build a statistical background for the use of Localized AUC for our purposes.
ER -