摘要
本文分别在Ⅱ型删失和随机删失下,表明了共轭先验下的指数分布的刻度参数的贝叶斯估计为具有如下形式的收缩估计(?)_(BE)=a■+bEθ,此处■为依赖样本θ的一个无偏估计且Eθ表示先验分布的期望。当采用平方损失函数时,a+b=1;如果用加权平方损失函数,则a+b<1。
Under the type Ⅱ censorship and random censorship, respectively, we show in this paper that the Bayes estimator of the exponential scale parameter with conjugate prior can be shrinkage estimation with the form θ^BE = αθ^ +b bEθ, where θ^ is an unbiased estimator depending on samples and Eθ denotes the expectation of the prior distribution. When the squared loss function is adopted, α + b = 1; if we use the weighted square loss filnction, then α+b〈 1.
出处
《工程数学学报》
CSCD
北大核心
2006年第3期553-558,共6页
Chinese Journal of Engineering Mathematics
基金
The National Natural Science Foundation of China(10271001)
关键词
贝叶斯估计
收缩估计
损失函数
共轭先验
Bayes estimation
shrinkage estimation
loss function
conjugate prior