Prior-based Bayesian information criterion

作     者：M.J.Bayarria James O.Berger Woncheol Jang Surajit Ray Luis R.Pericchi Ingmar Visser 

作者机构：Department of Statistics and Operations ResearchUniversity of ValenciaValenciaSpain Department of Statistical ScienceDuke UniversityDurhamNCUSA Department of StatisticsSeoul National UniversitySeoulKorea School of Mathematics and StatisticsUniversity of GlasgowGlasgowUK Department of MathematicsUniversity of Puerto RicoSan JuanPuerto Rico Department of PsychologyUniversity of AmsterdamAmsterdamNetherlands 

出 版 物：《Statistical Theory and Related Fields》 (统计理论及其应用（英文）)

年 卷 期：2019年第3卷第1期

页      面：2-13页

学科分类：07[理学] 0714[理学-统计学（可授理学、经济学学位）] 0701[理学-数学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 070101[理学-基础数学] 

基　　金：M.J.Bayarri’s research was supported by the Spanish Ministry of Education and Science[grant number MTM2010-19528] James Berger’s research was supported by USA National Science Foundation[grant numbers DMS-1007773 and DMS-1407775] Woncheol Jang’s research was supported by the National Research Foundation of Korea(NRF)grants funded by the Korea government(MSIP),No.2014R1A4A1007895 and No.2017R1A2B2012816 Luis Pericchi’s research was supported by grant CA096297/CA096300 from the USA National Cancer Institute of the National Institutes of Health 

主　　题：Bayes factors model selection Cauchy priors consistency effective sample size Fisher information Laplace expansions robust priors 

摘      要：We present a new approach to model selection and Bayes factor determination,based on Laplaceexpansions(as in BIC),which we call Prior-based Bayes Information Criterion(PBIC).In thisapproach,the Laplace expansion is only done with the likelihood function,and then a suitableprior distribution is chosen to allow exact computation of the(approximate)marginal likelihoodarising from the Laplace approximation and the *** result is a closed-form expression similar to BIC,but now involves a term arising from the prior distribution(which BIC ignores)andalso incorporates the idea that different parameters can have different effective sample sizes(whereas BIC only allows one overall sample size n).We also consider a modification of PBIC whichis more favourable to complex models.