指数分布定数截尾样本下经验Bayes双侧检验问题  

Empirical Bayes two-sided test problem based on type Ⅱ censored samples in the exponential distribution

作　　者：洪坚[1] 韦来生[1] 

机构地区：[1]中国科学技术大学统计与金融系,安徽合肥230026

出　　处：《中国科学技术大学学报》2006年第12期1289-1293,共5页JUSTC

基　　金：教育部博士点基金(20030358053);中国科学院知识创新工程重点方向项目(KJCX3-SYW-S02)资助

摘　　要：研究了指数分布定数截尾情形下失效率函数的经验Bayes(empirical Bayes,EB)双侧检验问题.利用概率密度函数的核估计构造了EB检验函数,证明了它的渐近最优性,并获得了其收敛速度.最后,给出了一个满足定理条件的例子.The empirical Bayes （EB） two-sided test problem of the failure function based on type Ⅱ censored samples in the exponential distribution was studied. The EB test rule was constructed by the kernel estimation of probability density function and its derivatives. Under suitable conditions, it is shown that the proposed EB test rules are asymptotically optimal with convergence rates near O（n^-1/2）. Finally, an example satisfying the conditions of the theorem was given.

关 键 词：指数分布 定数截尾 经验BAYES检验 渐近最优性 收敛速度 

分 类 号：O212.1[理学—概率论与数理统计]