LINEX损失下一类分布族参数渐近最优的EB检验

Asymptotically optimal empirical Bayes test for the parameter of two-sided truncated distribution family under LINEX loss function

目的 在LINEX损失函数下,讨论一类双边截断型分布族参数的经验Bayes(EB)检验问题。方法 构造适当的EB判决(检验)函数。结果 在经验Bayes检验问题中,将损失函数推广为LINEX损失,在适当的条件下证明了所构造的判决函数是渐近最优的。结论 LINEX损失函数具有比对称损失更广泛的意义,而且在一定条件下可以获得参数渐近最优的EB检验。

AimThe empirical Bayes test is considered for the parameter of two-sided truncated distribution family under LINEX loss.MethodsEB test function is constructed to be the basis.ResultsIn the EB test problem,loss function is generalized to LINEX loss,and it is shown that the constructed test function is asymptotically optimal.ConclusionLINEX loss function is more significant than symmetry loss,and asymptotically optimal EB test can be obtained under some proper assumption.