Comparison of Mixture and Classification Maximum Likelihood Approaches in Poisson Regression Models

Abstract

In this work, we propose to compare two algorithms to compute maximum likelihood estimators of the parameters of a mixture Poisson regression models. To estimate these parameters, we may use the EM algorithm in a mixture approach or the CEM algorithm in a classification approach. The comparison of the two procedures was done through a simulation study of the performance of these approaches on simulated data sets in a target number of iterations. Simulation results show that the CEM algorithm is a good alternative to the EM algorithm for fitting Poisson mixture regression models, having the advantage of converging more quickly.