摘要
本文结合贝叶斯先验和删失数据自回归模型的拟似然,构造此类模型下的贝叶斯变量选择,通过连续spike和slab先验和0-1潜变量,运用EM算法得出有效的参数估计和变量选择方法,该方法计算量小且模型正确识别率高。在模拟研究和实证分析中验证了此方法的有效性。
The Bayesian prior and quasi-likelihood of autoregressive model with censored data were combined to construct Bayesian variable selection in this paper.We used EM algorithm to derive effective parameter estimation and variable selection approach by applying continuous spike and slab prior distribution and 0-1 latent variables.This method has the advantages of small amount of calculation and high correct recognition rate of the model.The effectiveness of this method is verified in Simulation study and empirical analysis.
作者
袁晓惠
周世权
王岳
李会贤
YUAN Xiao-hui;ZHOU Shi-quan;WANG Yue;LI Hui-xian(School of mathematics and statistics,Changchun University of Technology,Changchun 130012,China)
出处
《数理统计与管理》
CSSCI
北大核心
2021年第5期833-841,共9页
Journal of Applied Statistics and Management
基金
国家自然科学基金(11671054,11571051).
