Comparison of Estimators of Dispersion Matrix

Abstract

Based on a sample, we considered the problem of estimating the dispersion matrix of a multivariate normal distribution with variance covariance matrix Σ. Empirical Bayes estimators and Haff estimators with their conditions, two proposed estimators of Σ, were the best affine equivariant estimators of dispersion matrix, which we compared them by three different loss functions.