分组数据场合逆威布尔分布参数贝叶斯估计的混合Gibbs算法

Mixed Gibbs algorithm of Bayesian estimation of parameters of inverse Weibull distribution in the grouped data occasions

作者利用混合Gibbs算法(Gibbs抽样与Metropolis算法的混合)给出了分组数据场合逆威布尔分布参数的贝叶斯估计,然后通过Monte-Carlo模拟考查了贝叶斯估计的均值、均方误差及参数的可信区间,并与极大似然估计比较,给出了混合Gibbs抽样过程中相应参数的轨迹图、直方图及自相关系数图.在五组分组数据场合用混合Gibbs算法求逆威布尔分布参数的贝叶斯估计都得到了比较满意的结果,表明该算法可行、稳定、并且有效.

This paper proposes Bayesian estimation of inverse Weibull distribution parameters using mixed Gibbs algorithm （Gibbs sampling with Metropolis algorithm mixed） in the grouped data occa- sions, examines the estimated mean, mean square error and the confidence interval for the parameter, gives and compares Bayesian estimation with maximum likelihood estimation by the Monte-Carlo simula- tion, and the locus,histograms and autocorrelation coefficient figure of the corresponding parameter in the the mixed Gibbs sampling process are obtained, Bayesian estimation of the inverse Weibull distribu- tion parameters using mixed Gibbs algorithm in the five grouped data occasions are quite satisfied. So the algorithm is a viable, stable and effective.