排序集抽样下Pareto分布形状参数的Bayes估计  

Bayesian Estimation of Pareto Distribution Shape Parameters Under Sorted Set Sampling

作　　者：宗凤喜 李如兵 Zong Fengxi;Li Rubing(School of Mathematics and Statistics, Qujing Normal University, Qujing Yunnan 655011, China;School of Economics, Shanghai University of Finance and Economics, Shanghai 200433, China)

机构地区：[1]曲靖师范学院数学与统计学院,云南曲靖655011 [2]上海财经大学经济学院,上海200433

出　　处：《统计与决策》2018年第24期15-19,共5页Statistics & Decision

基　　金：国家自然科学基金资助项目(11271013);云南省教育厅科学研究基金项目(2014C135Y;2017ZDX149)

摘　　要：文章在排续集抽样下,针对共轭先验以及Jeffreys先验,基于平方损失、Q-对称熵损失以及Linex损失获得了Pareto分布形状参数的Bayes估计。利用Monte Carlo法,计算出估计的偏差以及均方误差,并与在简单随机抽样下获得的相应值进行比较,模拟结果表明在排续集抽样下获得的估计更有效。通过Monte Carlo模拟证明,在相同条件下,基于共轭先验分布得到的估计比基于Jeffreys先验得到的估计更有效。Bayesian estimation of Pareto distribution shape parameters based on squared error loss function, Q-symmetric entropy loss function and Linex loss function are obtained for conjugate and Jeffreys prior distribution under ranked set sampling.This paper uses Monte Carlo method to calculate the estimated deviation and mean square error, and compares with the corresponding values obtained in simple random sampling. Simulation result shows that estimates obtained under the sequels sampling are more effective; and Monte Carlo simulation indicates that under the same conditions, the estimation based on conjugate prior distribution is more effective than that based on Jeffreys prior.

关 键 词：排序集抽样 共轭先验 Jeffreys先验 平方损失函数 Q-对称熵损失函数 LINEX损失函数 

分 类 号：O212.8[理学—概率论与数理统计]