摘要
提出对称熵损失函数的一般形式(λ/δ)q+(δ/λ)q-2(q>0),即q对称熵损失.讨论指数分布的尺度参数在此损失函数下的最小风险同变估计、Bayes估计和最小最大估计,给出了更具一般性的结论,并研究了(cT+d)-1形式估计的可容许性和不可容许性.
The present paper presents the general form of the symmetric entropy loss function, (λ/δ)~q+(δ/λ)~q-2(q>0), which is called the q-symmetric entropy loss function, and deals with the Byaesian estimation, the minimum risk equivalent estimation and the minimax estimation for the scale parameter of the (exponential) distribution under the loss function, the conclusion is generalized. The admissbility and inadmissibility of estimator with the form of (cT+d)^(-1) are discussed.
出处
《吉林大学学报（理学版）》
CAS
CSCD
北大核心
2005年第1期10-15,共6页
Journal of Jilin University:Science Edition
关键词
BAYES估计
同变估计
最小最大估计
尺度参数
可容许性
q对称熵损失函数
Bayesian estimation
equivalent estimation
minimax estimation
scale parameter
admissibility
q-symmetric entropy loss function
