KIT | KIT-Bibliothek | Impressum | Datenschutz

Stable reliability diagrams for probabilistic classifiers

Dimitriadis, Timo; Gneiting, Tilmann; Jordan, Alexander I.

Abstract:
A probability forecast or probabilistic classifier is reliable or calibrated if the predicted probabilities are matched by ex post observed frequencies, as examined visually in reliability diagrams. The classical binning and counting approach to plotting reliability diagrams has been hampered by a lack of stability under unavoidable, ad hoc implementation decisions. Here, we introduce the CORP approach, which generates provably statistically consistent, optimally binned, and reproducible reliability diagrams in an automated way. CORP is based on nonparametric isotonic regression and implemented via the pool-adjacent-violators (PAV) algorithm—essentially, the CORP reliability diagram shows the graph of the PAV-(re)calibrated forecast probabilities. The CORP approach allows for uncertainty quantification via either resampling techniques or asymptotic theory, furnishes a numerical measure of miscalibration, and provides a CORP-based Brier-score decomposition that generalizes to any proper scoring rule. We anticipate that judicious uses of the PAV algorithm yield improved tools for diagnostics and inference for a very wide range of statistical and machine learning methods.

Open Access Logo


Verlagsausgabe §
DOI: 10.5445/IR/1000130510
Veröffentlicht am 19.03.2021
Cover der Publikation
Zugehörige Institution(en) am KIT Institut für Stochastik (STOCH)
Publikationstyp Zeitschriftenaufsatz
Publikationsjahr 2021
Sprache Englisch
Identifikator ISSN: 0027-8424, 1091-6490
KITopen-ID: 1000130510
Erschienen in Proceedings of the National Academy of Sciences of the United States of America
Verlag National Academy of Sciences
Band 118
Heft 8
Schlagwörter calibration; discrimination ability; probability forecast; score decomposition; weather prediction
Nachgewiesen in Web of Science
Scopus
KIT – Die Forschungsuniversität in der Helmholtz-Gemeinschaft
KITopen Landing Page