随机搜索变量方法在滞后回归模型中的应用  

The Applications of Stochastic Search Variable Selection in Lag Regressive Model

作　　者：曾丽娟[1] 

机构地区：[1]中山大学新华学院,广东广州510642

出　　处：《杭州师范大学学报（自然科学版）》2014年第6期664-668,共5页Journal of Hangzhou Normal University(Natural Science Edition)

摘　　要：对滞后回归模型进行贝叶斯分析,并将随机搜索选择变量方法引入模型中,基于香港空气污染和患呼吸疾病人数数据建立时间序列滞后回归模型.构造基于Gibbs抽样的MCMC数值计算对该模型进行变量选择和参数估计,分析结果表明:湿度对香港患呼吸疾病人数影响最显著,二氧化氮和悬浮粒子也是影响比较显著的空气污染物,模型体现了香港空气污染的真实情况.This paper makes the Bayesian analysis on the lag regressive model,introduces Stochastic Search Variable Selection into the model,and constructs a time series lag regressive model based on the air pollution and respiratory disease patients data in Hongkong.Using the numerical calculation of MCMC based on Gibbs sampling to make variable selection and parameter estimation on the model,the analysis results point out that humidity has the most significant effects on the number of the patients with respiratory disease,NO2 and TSP particles also are the air pollutants with great effects.The model reflects the real situation of the air pollution in Hongkong.

关 键 词：滞后回归模型 随机搜索选择变量 香港空气污染 GIBBS抽样 

分 类 号：O212.8[理学—概率论与数理统计]