KIT | KIT-Bibliothek | Impressum | Datenschutz

Receiver operating characteristic (ROC) curves: equivalences, beta model, and minimum distance estimation

Gneiting, Tilmann 1; Vogel, Peter
1 Karlsruher Institut für Technologie (KIT)

Abstract:

Receiver operating characteristic (ROC) curves are used ubiquitously to evaluate scores, features, covariates or markers as potential predictors in binary problems. We characterize ROC curves from a probabilistic perspective and establish an equivalence between ROC curves and cumulative distribution functions (CDFs). These results support a subtle shift of paradigms in the statistical modelling of ROC curves, which we view as curve fitting. We propose the flexible two-parameter beta family for fitting CDFs to empirical ROC curves and derive the large sample distribution of minimum distance estimators in general parametric settings. In a range of empirical examples the beta family fits better than the classical binormal model, particularly under the vital constraint of the fitted curve being concave.


Verlagsausgabe §
DOI: 10.5445/IR/1000143005
Veröffentlicht am 13.02.2022
Originalveröffentlichung
DOI: 10.1007/s10994-021-06115-2
Scopus
Zitationen: 9
Dimensions
Zitationen: 14
Cover der Publikation
Zugehörige Institution(en) am KIT Fakultät für Mathematik (MATH)
Publikationstyp Zeitschriftenaufsatz
Publikationsjahr 2022
Sprache Englisch
Identifikator ISSN: 0885-6125, 1573-0565
KITopen-ID: 1000143005
Erschienen in Machine learning
Verlag Springer-Verlag
Band 111
Seiten 2147–2159
Vorab online veröffentlicht am 17.12.2021
Schlagwörter Binary prediction, Classification, Evaluation of predictive potential
Nachgewiesen in Dimensions
Scopus
Web of Science
KIT – Die Forschungsuniversität in der Helmholtz-Gemeinschaft
KITopen Landing Page