A survey of cross-validation procedures for model selection

被引：2669

作者：

Arlot, Sylvain ^{[1
]}

Celisse, Alain ^{[2
]}

机构：

[1] CNRS, Willow Project Team, Lab Informat, CNRS ENS INRIA UMR 8548,Ecole Normale Super, 23 Ave Italie, F-75214 Paris 13, France

[2] Univ Lille 1, CNRS, UMR 8524, Lab Math Paul Painleve, F-59655 Villeneuve, France

来源：

STATISTICS SURVEYS | 2010年 / 4卷

关键词：

Model selection; cross-validation; leave-one-out;

D O I：

10.1214/09-SS054

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：

Used to estimate the risk of an estimator or to perform model selection, cross-validation is a widespread strategy because of its simplicity and its (apparent) universality. Many results exist on model selection performances of cross-validation procedures. This survey intends to relate these results to the most recent advances of model selection theory, with a particular emphasis on distinguishing empirical statements from rigorous theoretical results. As a conclusion, guidelines are provided for choosing the best cross-validation procedure according to the particular features of the problem in hand.

引用

页码：40 / 79

页数：40

共 125 条

[1] STATISTICAL PREDICTOR IDENTIFICATION [J].