An Empirical Bayes Estimate of Multinomial Probabilities.

摘要

This paper deals with the problem of estimating simultaneously the parameters (cell probabilities) of m = or > 2 independent multinomial distributions, with respect to a quadratic loss function. An empirical Bayes estimator is given which is shown to have smaller risk than the maximum likelihood estimator for sufficiently large value of m, at all points of the parameter space outside its boundary. (Author)