BAYESIAN AND FREQUENTIST APPROACHES FOR FITTING THE GAMMA-TIME-DEPENDENT MODEL TO DESCRIBE NEUTRAL DETERGENT FIBER DEGRADATION

Abstract

The aim of the study was evaluate and compare the efficiency of Bayesian and frequentist approach to describe the rumen degradation of NDF. Simulated data was composed by four scenarios: regular restriction in the number of incubation times, random loss of incubation times, loss of specific parts of degradation curves, variation in the precision of the incubations procedures. Two real datasets was used, these real data encompassed the evaluation of NDF degradation of a tropical grass (Brachiaria decumbes).  The model was fitted according their characteristics approach and compared by plots and assessors. The Bayesian and frequentist approach presented reliable estimates of degradation parameters for the majority of the data tested. Therefore, in specific cases with short random records number, the Bayesian approach showed greater bias of the estimates of incubation residue and estimates  of degradation rate without a biological coherence of the parameters, compared to frequentist  inference. In another words, the Bayesian approach fitted with prior diffuse, presented less flexible. Nevertheless, it is emphasized the importance of the background information before the modeling, mainly  for the Bayesian approach, in order to define proper prior distributions. Future thorough studies about the influence of non-informative prior for the parameters are necessary.