ABSTRACT
In the Bayesian and Non-Bayesian statistical modeling, we are naturally involved in the question of model selection. In the Bayesian linear regression modeling in Section 2.7, we studied that it is important to check how independent variables affect a response variable of interest. In other words, we have to select a set of variables that is important to predict the response variable. Also, the prior setting affected the prediction results. Although we assumed the normal error term for a simplicity of illustration of Bayesian regression modeling, we can also employ fat-tailed error terms, e.g., Student-t sampling density. The estimated Bayesian model depends on the specifications of the sampling density structure and the prior distribution of the model parameters, and thus crucial issues with Bayesian statistical modeling are the model evaluation.
