Comparison Between Maximum Likelihood and Bayesian Methods For Estimating The Gamma Regression With Practical Application

Abstract

In this paper, we will illustrate a gamma regression model assuming that the dependent variable (Y) is a gamma distribution and that it's mean ( ) is related through a linear predictor with link function which is identity link function g(μ) = μ. It also contains the shape parameter which is not constant and depends on the linear predictor and with link function which is the log link and we will estimate the parameters of gamma regression by using two estimation methods which are The Maximum Likelihood and the Bayesian and a comparison between these methods by using the standard comparison of average squares of error (MSE), where the two methods were applied to real data on the disease of jaundice of children newborns(Infant Jaundice) and it was the best method of estimation It is the Maximum Likelihood because it gave less (MSE).