We begin by reviewing the Bayesian formulation of a linear model and show that instead of point estimates for parameters and predictions, we obtain an entire posterior and predictive distribution.
