An EM Algorithm for Multivariate Mixed Poisson Regression Models and its Application
Mohamed E. Ghitany, Dimitris Karlis, Fahimah A. Al-Awadhi · 2012
Although the literature on univariate count regression models allowing for overdispersion is huge, there are few multivariate count regression models allowing for correlation and overdiseprsion. The latter models can find applications in several disciplines such as epidemiology, marketing, sports statistics, criminology, just to name a few. In this paper, we propose a general EM algorithm to facilitate maximum likelihood estimation for a class of multivariate mixed Poisson regression models. We give special emphasis to the multivariate negative binomial, Poisson inverse Gaussian and Poisson lognormal regression models. An application to a real dataset is also given to illustrate the use of the proposed EM algorithm to the considered multivariate regression models. Mathematics Subject Classification: 62J05, 65D15