Asymptotic efficient semiparametric empirical bayes estimation of multinomial responses

Abstract

In this paper, Bayesian statistical inference for estimating of parameters in a Semi-parametric Multinomial Logit Models with repeated experiments is developed. Our method is based on empirical Bayes estimation. The results are shown, estimators are consistent and asymptotically efficient. Such approach allows us to take a new accurate theory using reliable and quick iteration algorithm to compute the empirical Bayes estimate. Results of real examples show that the proposed method is very competitive in terms of estimation accuracy and speed of computational estimation methods.

Keywords:

• Bayes
• Bayesian Nonparametric
• Finite Mixture model
• Multinomial Distribution
• Model Selection