We study improved shrinkage estimation of a vector of non-negative means. We concentrate on the Gaussian case with known scale, but do not necessarily assume the initial estimator is minimax. As a result, we find improved shrinkage estimators in fewer than 3 dimension in certain cases. Generalized Bayes estimators which may be improved via shrinkage in 1 and 2 dimensions illustrate the result. We also consider improved positive part estimators.
All Science Journal Classification (ASJC) codes
- Statistics and Probability
- Statistics, Probability and Uncertainty
- Generalized Bayes estimator
- Katz estimator
- Pseudo-Bayes estimator
- Stein estimator