Stable reliability diagrams for probabilistic classifiers

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 l...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Dimitriadis, Timo (VerfasserIn) , Gneiting, Tilmann (VerfasserIn) , Jordan, Alexander I. (VerfasserIn)
Dokumenttyp: Article (Journal)
Sprache:Englisch
Veröffentlicht: February 17, 2021
In: Proceedings of the National Academy of Sciences of the United States of America
Year: 2021, Jahrgang: 118, Heft: 8, Pages: 1-10
ISSN:1091-6490
DOI:10.1073/pnas.2016191118
Online-Zugang:Verlag, lizenzpflichtig, Volltext: https://doi.org/10.1073/pnas.2016191118
Verlag, lizenzpflichtig, Volltext: https://www.pnas.org/content/118/8/e2016191118
Volltext
Verfasserangaben:Timo Dimitriadis, Tilmann Gneiting, and Alexander I. Jordan

MARC

LEADER 00000caa a2200000 c 4500
001 1753154731
003 DE-627
005 20220819154241.0
007 cr uuu---uuuuu
008 210406s2021 xx |||||o 00| ||eng c
024 7 |a 10.1073/pnas.2016191118  |2 doi 
035 |a (DE-627)1753154731 
035 |a (DE-599)KXP1753154731 
035 |a (OCoLC)1341403386 
040 |a DE-627  |b ger  |c DE-627  |e rda 
041 |a eng 
084 |a 17  |2 sdnb 
100 1 |a Dimitriadis, Timo  |e VerfasserIn  |0 (DE-588)1230883045  |0 (DE-627)1753224217  |4 aut 
245 1 0 |a Stable reliability diagrams for probabilistic classifiers  |c Timo Dimitriadis, Tilmann Gneiting, and Alexander I. Jordan 
264 1 |c February 17, 2021 
300 |a 10 
336 |a Text  |b txt  |2 rdacontent 
337 |a Computermedien  |b c  |2 rdamedia 
338 |a Online-Ressource  |b cr  |2 rdacarrier 
500 |a Gesehen am 07.04.2021 
520 |a 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. 
650 4 |a calibration 
650 4 |a discrimination ability 
650 4 |a probability forecast 
650 4 |a score decomposition 
650 4 |a weather prediction 
700 1 |a Gneiting, Tilmann  |e VerfasserIn  |0 (DE-588)1019627484  |0 (DE-627)690974809  |0 (DE-576)358470323  |4 aut 
700 1 |a Jordan, Alexander I.  |e VerfasserIn  |0 (DE-588)1027203264  |0 (DE-627)72857747X  |0 (DE-576)372589928  |4 aut 
773 0 8 |i Enthalten in  |a National Academy of Sciences (Washington, DC)  |t Proceedings of the National Academy of Sciences of the United States of America  |d Washington, DC : National Acad. of Sciences, 1915  |g 118(2021), 8, Artikel-ID e2016191118, Seite 1-10  |h Online-Ressource  |w (DE-627)254235379  |w (DE-600)1461794-8  |w (DE-576)073260509  |x 1091-6490  |7 nnas 
773 1 8 |g volume:118  |g year:2021  |g number:8  |g elocationid:e2016191118  |g pages:1-10  |g extent:10  |a Stable reliability diagrams for probabilistic classifiers 
856 4 0 |u https://doi.org/10.1073/pnas.2016191118  |x Verlag  |x Resolving-System  |z lizenzpflichtig  |3 Volltext 
856 4 0 |u https://www.pnas.org/content/118/8/e2016191118  |x Verlag  |z lizenzpflichtig  |3 Volltext 
951 |a AR 
992 |a 20210406 
993 |a Article 
994 |a 2021 
998 |g 1027203264  |a Jordan, Alexander  |m 1027203264:Jordan, Alexander  |p 3  |y j 
998 |g 1019627484  |a Gneiting, Tilmann  |m 1019627484:Gneiting, Tilmann  |p 2 
998 |g 1230883045  |a Dimitriadis, Timo  |m 1230883045:Dimitriadis, Timo  |d 180000  |d 181000  |e 180000PD1230883045  |e 181000PD1230883045  |k 0/180000/  |k 1/180000/181000/  |p 1  |x j 
999 |a KXP-PPN1753154731  |e 3902111550 
BIB |a Y 
SER |a journal 
JSO |a {"physDesc":[{"extent":"10 S."}],"relHost":[{"part":{"extent":"10","text":"118(2021), 8, Artikel-ID e2016191118, Seite 1-10","volume":"118","issue":"8","pages":"1-10","year":"2021"},"titleAlt":[{"title":"PNAS online"},{"title":"PNAS"}],"pubHistory":["1.1915 -"],"recId":"254235379","corporate":[{"roleDisplay":"VerfasserIn","display":"National Academy of Sciences","role":"aut"}],"language":["eng"],"disp":"National Academy of Sciences (Washington, DC)Proceedings of the National Academy of Sciences of the United States of America","note":["Gesehen am 09.03.2022"],"type":{"media":"Online-Ressource","bibl":"periodical"},"title":[{"title":"Proceedings of the National Academy of Sciences of the United States of America","title_sort":"Proceedings of the National Academy of Sciences of the United States of America"}],"physDesc":[{"extent":"Online-Ressource"}],"id":{"issn":["1091-6490"],"eki":["254235379"],"zdb":["1461794-8"]},"origin":[{"publisherPlace":"Washington, DC","publisher":"National Acad. of Sciences","dateIssuedKey":"1915","dateIssuedDisp":"1915-"}]}],"origin":[{"dateIssuedDisp":"February 17, 2021","dateIssuedKey":"2021"}],"id":{"doi":["10.1073/pnas.2016191118"],"eki":["1753154731"]},"name":{"displayForm":["Timo Dimitriadis, Tilmann Gneiting, and Alexander I. Jordan"]},"note":["Gesehen am 07.04.2021"],"type":{"media":"Online-Ressource","bibl":"article-journal"},"language":["eng"],"recId":"1753154731","title":[{"title_sort":"Stable reliability diagrams for probabilistic classifiers","title":"Stable reliability diagrams for probabilistic classifiers"}],"person":[{"family":"Dimitriadis","given":"Timo","roleDisplay":"VerfasserIn","display":"Dimitriadis, Timo","role":"aut"},{"role":"aut","roleDisplay":"VerfasserIn","display":"Gneiting, Tilmann","given":"Tilmann","family":"Gneiting"},{"role":"aut","display":"Jordan, Alexander I.","roleDisplay":"VerfasserIn","given":"Alexander I.","family":"Jordan"}]} 
SRT |a DIMITRIADISTABLERELI1720