This research presents and assesses a novel model for time-series count data, the Mixture Poisson Autoregressive (MPAR) model, which addresses the challenges of discreteness, overdispersion, and serial correlation. The parametric technique is used for the MPAR model with a marginal distribution for the count data. The MPAR model was applied to Anaemia cases among exclusive and non-exclusive breastfeeding children in Mubi North Local Government Area. The results show a multimodal distribution of cases, thus revealing that the disease is a product of several risk factors. The MPAR model with two components provided a good fit to the data.
Download