左截断右删失数据下指数分布多变点Bayes估计

his paper is based on left truncated and right censored data. The complete-data likelihood function of exponential distribution is obtained after the data input. The fisher information matrix is used to determine each priori parameter information, and the full conditional distribution with multiple change points of each parameter is deduced. The MCMC method of Gibbs sampling together with Metropolis-Hastings algorithm is used to get the samples. Finally, random simulation is done by R software. The test result indicates that using the Bayes method to handle the exponential distribution with Multiple Change Points for left truncated and right censored data is effective, in which the relative error of estimations of change-point position parameters is less than 3%.