Generalized Linear Models

This chapter presents generalized linear models, including the maximum likelihood and Bayesian methods used to obtain their parameter inferences. Starting from the one-parameter exponential family, the chapter defines the generalized linear models and discusses the maximum likelihood methods used to obtain their parameter estimates, including the Newton-Raphson and the Fisher scoring algorithms. After that, generalized Bayesian models are defined, including the Gibbs sampling and the Metropolis-Hasting algorithms used to propose the Bayesian methods employed to obtain the posterior parameter inferences. Some examples illustrate the models&s; applications.