In this paper we discuss the scale parameter estimation problem of the two-parameter exponential distribution given location parameter under the typeⅡcensorship and random censorship. We show the Bayes estimator of the scale parameter with conjugate prior can be shrinkage estimation with the formθBE=aθ+bEθ, whereθis an unbiased estimator depending on samples and Eθis the expectation of the prior distribution g(θ).When we use the squared loss function, a+b=1.When we use the weighted squared loss function and r=1,then a+b=1; if r>1,then a+b<1. Comparing the two kinds of estimators, we know that the Bayes estimator is better than all the shrinkage estimators with form p=k(?)+(1-k)Eθ.
|