The Lasso regression model is one of the simplest and most popular penalized regression models in statistics. In this talk we describe a simple Monte Carlo method for sampling from the posterior density of the Bayesian posterior of the Lasso. We show how the construction of such an efficient Monte Carlo sampler necessitates that we solve the frequentist Lasso optimization problem in a novel way. We give a numerical example with the well-known diabetes dataset of Efron.
This is joint work with Pierre L'Ecuyer.
Welcome to everyone!