逐步二型删失数据下广义Pareto分布的参数估计  

Parameter estimation of generalized Pareto distributionunder progressively type-II censored data

作　　者：苏燕青 金良琼 邹路燕 李琼忆 陶永 SU Yanqing;JIN Liangqiong;ZOU Luyan;LI Qiongyi;Tao Yong(College of Data Science and Information Engineering,Guizhou Minzu University,Guiyang,Guizhou 550025,China)

机构地区：[1]贵州民族大学数据科学与信息工程学院,贵州贵阳550025

出　　处：《内江师范学院学报》2024年第4期45-49,共5页Journal of Neijiang Normal University

基　　金：贵州省科技计划基金项目(黔科合基础-ZK[2021]一般342)。

摘　　要：逐步二型删失数据在生存分析中的应用较为广泛,因此在逐步二型删失数据下推导了广义Pareto分布中的尺度参数的极大似然估计与Bayes估计的估计形式,并在平方损失函数下,基于MCMC方法与Lindley近似法,给出了广义Pareto分布尺度参数的Bayes估计.结果表明:贝叶斯估计方法优于极大似然估计方法,MCMC后验抽样下的Bayes估计优于Lindley近似方法.Progressively type-II censored data is widely used in survival analysis.Therefore,the estimation forms of maximum likelihood estimation and bayes estimation for scale parameters in generalized Pareto distribution are derived under progressively type-ii censored data.Under the square loss function,based on MCMC method and Lindley approximation method,the Bayes estimation of generalized Pareto distribution scale parameters is given.The results show that Bayesian estimation is better than maximum likelihood estimation,and Bayesian estimation under MCMC posterior sampling is better than Lindley approximation.

关 键 词：广义PARETO分布 Lindley近似法 BAYES估计 极大似然估计 M-H算法 

分 类 号：O213.2[理学—概率论与数理统计]