Abordagem Bayesiana para estimadores de encolhimento

Abstract

The statistician-mathematician Charles Stein, in 1955, in a publication “Inadmissibility of the usual estimator for the mean of a multivariate normal distribution" (GRUBER, 1998) surprised the world of statistics with its proof that the maximum likelihood estimator is inadmissible, except in the one-dimensional and two-dimensional cases. Stein showed that, in case a biased estimator is admitted, there are estimators with mean square error inferior to the mean square error of the maximum likelihood estimator. These estimators comprise a class denominated shrinkage estimators. These estimators have, in general, mean square error lower than the usual estimators, as will be presented over the course of this work.