Nonparametric empirical bayes estimation of the distribution function and the mean
Malay Ghosh, Parthasarthi Lahiri, Ram Chandra Tiwari · Communication in Statistics- Theory and Methods · 1989
The paper considers nonparametric empirical Bayes estimation of the distribution function and the mean both for finite and infinite populations. The proposed estimators are optimal within the class of estimators which are weighted averages of empirical distribution functions when the underlying prior is a Dirichlet process prior. Comparison with certain rival estimators is made. The results are also obtained for estimable parameters of degree one with a Dirichlet invariant prior.