Bayesian modeling used to describe the drying kinetics of rice grains

Abstract

Rice is a cereal widely consumed throughout the world, it is a seasonal product, which justifies the need for conservation and storage. In the South American context, Brazil stands out as the main producer. The practice of
drying is highly beneficial, not only for minimizing post-harvest losses, but also for adding additional value to the final product. Studies generally employ few observations, but inference in nonlinear models is only valid for sufficiently large samples. As the Bayesian methodology has proven to be efficient even with small samples, the objective of this work is to analyze the adjustment of nonlinear regression models to rice grain drying kinetics data using the Bayesian inference approach. using informative and non-informative priors, as well as developing complete conditional distributions for the parameters of the evaluated models, samples of posterior marginal distributions were generated using the Gibbs sampler and Metropolis-Hastings algorithms implemented in the R software. The evaluated models were Henderson and Pabis and Simple Exponential with Three Parameters, obtaining the convergence diagnosis, mean and interval point estimates (HPD), as well as selection criteria such as the Bayesian Information Criteria (BIC) and Deviance (DIC), measure of Kullback-Leibler and Bayes factor. According to these criteria, it can be concluded that the Henderson and Pabis model was the most suitable for describing the drying kinetics of rice grains.