Implementation of generalized additive models for spatial beta regression

Abstract

Beta regression models are proposed by to model the continuous variates that assume values in the standard unit interval, e.g. rates, proportions or concentration, or inequality indices. These models belong to the class of generalized linear mixed models (GLM) or more general generalized additive models (GAM) with responses belonging to the exponential family. In present study we use GAM to model spatial Beta data. We develop the Monte Carlo version of EM algorithm for obtaining of penalized maximum likelihood estimators of model parameters. This method is applied to real data set on Black carrageen concentration index to obtain a model of its distribution over the southeastern Baltic Sea