截断删失情形下泊松分布参数多变点的Bayes估计

Bayes Estimation of Parameter of Poisson Distribution with Multiple Change Points for Randomly Truncated and Censored Data

通过添加数据得到截断删失情形下泊松分布的完全数据似然函数,研究变点位置和其它参数的满条件分布.利用Gibbs抽样与Metropolis-Hastings算法相结合的MCMC方法对参数进行估计,详细介绍MCMC方法的实施步骤.随机模拟试验的结果表明参数Bayes估计的精度较高.

By filling in the data,the complete-data likelihood function of Poisson distribution for truncated and censored data is obtained.The full conditional distributions of changepoint positions and other parameters are studied.The parameters are estimated by MCMC method of Gibbs sampling together with Metropolis-Hastings algorithm.The implementation steps of MCMC method are introduced in detail.The random simulation test results show that Bayes estimations of the parameters are fairly accurate.