Statistical Estimation by the Empirical Bayes Method: Some Extensions and Logistical Applications

摘要

This report contains a presentation of the empirical Bayes procedure for improving existing methods of estimating statistical decision parameters which occur in such logistical problems as reliability, maintenance, and supply. It consists essentially of estimating a parameter by using an approximate Bayes estimator that does not depend on any previous information, other than earlier observations. The author presents estimators for the univariate and multivariate exponential family of distributions, for distributions with nuisance parameters, and for the distribution of a family of random variables (Poisson process). (Author)