Chapter 11 Bayesian model averaging in variable selection

We will show how to perform Bayesian model averaging. There are two issue to handle here: models space and the marginal likelihood. The former will be tackled using Markov chain Monte Carlo model composition, Occam’s window and stochastic search variable selection. The latter can be overcome using analytical solutions or Laplace approximation. We will show the theory underlying these approaches, how to perform BMA using our GUI and R for linear models: exogenous and endogenous, and non-linear models: logit, gamma and poisson. We will have also mathematical and computational exercises in R and our GUI.