1篇
您的检索式：作者名="Ruobin Gong"
|
|
|
题名
|
作者
|
年代
|
出处
|
被引量
|
| 1 | Discussion on Prior-based Bayesian Information Criterion(PBIC)by M.J.Bayarri,James O.Berger,Woncheol Jang,Surajit Ray,Luis R.Pericchi,andIngmar Visser显示文摘It is our pleasure to comment on this very interestingarticle on model selection.The Bayesian InformationCriterion(BIC)is one of the most popular metrics formodel selection.It is taught in every classroom on statistical modelling,and widely implemented as part of astandard routine in statistical programming languages.The fact that BIC is so popular means it is often erroneously applied to model classes that are beyond itsdesign.In this elucidating paper,the authors unpackeda dangerous complication when one takes the classicBIC verbatim as an approximation to the marginal likelihood.An additive constantc,ignored in the derivationof BIC,is in fact model dependent.When two BICs arecomputed on different models and their difference orratio used for comparison,their respective constants arealso implicitly factored into the comparison. | Ruobin Gong Minge Xie | 2019 | Statistical Theory and Related Fields2019，3，1： | 0 |
您所在的机构暂未开通此部分的功能，请联系管理员