Regular conditional probabilities and strictly proper loss functions

被引：1

作者：

Nielsen, Michael ^{[1
]}

机构：

[1] Univ Sydney, Sydney, NSW, Australia

来源：

STATISTICS & PROBABILITY LETTERS | 2022年 / 185卷

关键词：

Regular conditional probability; Strictly proper loss functions; Expected loss; Strict propriety;

D O I：

10.1016/j.spl.2022.109412

中图分类号：

O21 [概率论与数理统计]; C8 [统计学];

学科分类号：

020208 ; 070103 ; 0714 ;

摘要：

The main result of this paper is that, under fairly general conditions, regular conditional probabilities uniquely minimize the expected loss of a strictly proper loss function. A trivial converse to this result also allows us to characterize strict property in a purely probabilistic way. (c) 2022 Elsevier B.V. All rights reserved.

引用

页数：5

共 11 条

[1] On the optimality of conditional expectation as a Bregman predictor
Banerjee, A
Guo, X
Wang, H
[J]. IEEE TRANSACTIONS ON INFORMATION THEORY, 2005, 51 (07) : 2664 - 2669
[2] de Finetti Bruno., 1974, THEORY PROBABILITY C, VVolume 1
[3] THE EXISTENCE OF REGULAR CONDITIONAL PROBABILITIES - NECESSARY AND SUFFICIENT CONDITIONS
FADEN, AM
[J]. ANNALS OF PROBABILITY, 1985, 13 (01) : 288 - 298
[4] Strictly proper scoring rules, prediction, and estimation
Gneiting, Tilmann
Raftery, Adrian E.
[J]. JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 2007, 102 (477) : 359 - 378
[5] Justifying conditionalization: Conditionalization maximizes expected epistemic utility
Greaves, Hilary
Wallace, David
[J]. MIND, 2006, 115 (459) : 607 - 632
[6] KallenbergO, 2006, FDN MODERN PROBABILI
[7] Newman J, 2015, ARXIV PREPRINT ARXIV
[8] Ramsey Frank P. R., 1926, Philosophical Papers, P52
[9] Savage L. J., 1954, FDN STAT
[10] Seidenfeld T, 2001, ANN PROBAB, V29, P1612

← 1 2 →