序约束下单向分类方差分析模型的Bayes变量选择  

Bayesian Variable Selection in One-Way ClassifiedANOVA Models under Order Constraints

作　　者：史海芳[1] 李聪[2] 姬永刚[1] SHI Haifang;LI Cong;JI Yonggang(School of Science,Civil Aviation University of China,Tianjin 300300,China;College of Mathematics,Jilin University,Changchun 130012,China)

机构地区：[1]中国民航大学理学院,天津300300 [2]吉林大学数学学院,长春130012

出　　处：《吉林大学学报（理学版）》2021年第5期1093-1100,共8页Journal of Jilin University:Science Edition

基　　金：国家自然科学基金青年科学基金(批准号:12001229);中央高校基本科研业务费专项基金(批准号:3122014K013).

摘　　要：考虑序约束下单向分类方差分析模型的变量选择问题,提出两种基于SSVS的Bayes变量选择方法,并设计一种简单且易操作的Gibbs抽样算法进行后验抽样.数值模拟和应用实例结果表明,该方法效果较好.We considered variable selection problem in one-way classified ANOVA under order constraints.We proposed two Bayesian variable selection method based on the stochastic search variable selection(SSVS),and designed a simple and easy to operate Gibbs sampling algorithm for a posteriors sampling.The results of numerical simulation and application examples show that the proposed method is effective.

关 键 词：Bayes变量选择 序约束 随机搜索变量选择 

分 类 号：O212.1[理学—概率论与数理统计]