统计研究
統計研究
통계연구
Statistical Research
2014年
5期
102~106
,共null页
Bayes估计 Metropolis-Hastings MCMC
Bayes估計 Metropolis-Hastings MCMC
Bayes고계 Metropolis-Hastings MCMC
Bayesian Estimation; Metropolis-Hastings; MCMC
本文采用Bayes方法对空间滞后模型进行全面分析。在构建模型的贝叶斯框架时,对模型系数与误差方差分别选取正态先验分布和逆伽玛先验分布,以便获得参数的联合后验分布和条件后验分布。在抽样估计时,主要使用MCMC方法,同时还设计了一个简单随机游动Metropolis抽样器,以便从空间权重因子系数的条件后验分布中进行抽样。最后应用所建议的方法进行数值模拟。
本文採用Bayes方法對空間滯後模型進行全麵分析。在構建模型的貝葉斯框架時,對模型繫數與誤差方差分彆選取正態先驗分佈和逆伽瑪先驗分佈,以便穫得參數的聯閤後驗分佈和條件後驗分佈。在抽樣估計時,主要使用MCMC方法,同時還設計瞭一箇簡單隨機遊動Metropolis抽樣器,以便從空間權重因子繫數的條件後驗分佈中進行抽樣。最後應用所建議的方法進行數值模擬。
본문채용Bayes방법대공간체후모형진행전면분석。재구건모형적패협사광가시,대모형계수여오차방차분별선취정태선험분포화역가마선험분포,이편획득삼수적연합후험분포화조건후험분포。재추양고계시,주요사용MCMC방법,동시환설계료일개간단수궤유동Metropolis추양기,이편종공간권중인자계수적조건후험분포중진행추양。최후응용소건의적방법진행수치모의。
This paper used Bayesian method to analyze the spatial lag model. When the construction of Bayesian framework, we selected normal distribution and inverse gamma distribution as prior of model coefficients and error variance in order to obtain the joint posterior distribution and the conditional posterior distribution of parameters. The article mainly used MCMC methods to estimate parameters, and also designed a simple random walk Metropolis sampling device, to facilitate the sampling from conditional posterior distribution of spatial weighting factor coefficient. Finally, the proposed method was used for numerical simulation.