Bayesian Mixture Poisson Regression for Modeling Spatial Point Pattern of Primary Health Centers in Surabaya

Abstract

Primary Health Centres (PHC) are the first referral health facilities for Indonesian people to seek treatment. The varying distribution of PHC location in central Surabaya, therefore, causes its process to follow the Non-homogeneous Poisson Point Process (NHPP). We use Bayesian analysis coupled with Markov Chain Monte Carlo (MCMC) to model the mixture Poisson regression on NHPP intensity estimation. The result shows that two mixture components are significantly involved in the model along with four variables; i.e., the total population, the number of clean households, Accessibility Index, and the length of road; that produce the smallest Deviance Information Criteria (DIC).  Keywords: Bayesian Analysis; Mixture Poisson Regression; NHPP Intensity; Primary Health Centres