Semiparametric zero-inflated Poisson models with application to animal abundance studies

This paper describes a framework for flexibly modeling zero-inflated data. Semiparametric regression based on penalized regression splines for zero-inflated Poisson models is introduced. Moreover, an EM-type algorithm is developed to perform maximum likelihood estimation. As an illustration, a study of animal abundance is tackled. In fact, abundance often shows excess of zeroes and is a complicated function of the explanatory variables. In particular, the relationships between avian abundance and environmental variables indicating land use are tackled.