Empirical Bayes Estimation of the Multivariate Normal Covariance Matrix
Haff, L. R.
Ann. Statist., Tome 8 (1980) no. 1, p. 586-597 / Harvested from Project Euclid
Let $\mathbf{S}_{p \times p}$ have a Wishart distribution with scale matrix $\Sigma$ and $k$ degrees of freedom. Estimators of $\Sigma$ are given for each of the loss functions $L_1(\hat{\Sigma}, \Sigma) = \operatorname{tr} (\hat{\Sigma}\Sigma^{-1}) - \log \det (\hat{\Sigma}\Sigma^{-1}) - p$ and $L_2(\hat{\Sigma}, \Sigma) = \operatorname{tr} (\hat{\Sigma}\Sigma^{-1} - I)^2$. The obvious estimators of $\Sigma$ are the scalar multiples of $\mathbf{S}$, i.e., $a\mathbf{S}$ where $0 < a \leqslant 1/k$. (Recall that $(1/k)\mathbf{S}$ is unbiased.) For each problem $(\Sigma, \hat{\Sigma}, L_i), i = 1, 2$, we provide empirical Bayes estimators which dominate $a\mathbf{S}$ by a substantial amount. It is seen that the uniform reduction in the risk function determined by $L_2$ is at least $100(p + 1)/(k + p + 1){\tt\%}$. Dominance results for $L_1$ and $L_2$ were first given by James and Stein.
Publié le : 1980-05-14
Classification:  Covariance matrix,  empirical Bayes estimators,  unbiased estimation of risk function,  62F10,  62C99
@article{1176345010,
     author = {Haff, L. R.},
     title = {Empirical Bayes Estimation of the Multivariate Normal Covariance Matrix},
     journal = {Ann. Statist.},
     volume = {8},
     number = {1},
     year = {1980},
     pages = { 586-597},
     language = {en},
     url = {http://dml.mathdoc.fr/item/1176345010}
}
Haff, L. R. Empirical Bayes Estimation of the Multivariate Normal Covariance Matrix. Ann. Statist., Tome 8 (1980) no. 1, pp.  586-597. http://gdmltest.u-ga.fr/item/1176345010/