JSM 2015 Preliminary Program

Online Program Home
My Program

Abstract Details

Activity Number: 29
Type: Contributed
Date/Time: Sunday, August 9, 2015 : 2:00 PM to 3:50 PM
Sponsor: Section on Bayesian Statistical Science
Abstract #316779
Title: Standard Errors and Selection of Tuning Parameters for Bayesian Lassos Using Geometrically Ergodic Gibbs Samplers
Author(s): Sounak Chakraborty*
Companies: University of Missouri - Columbia
Keywords: Bayesian lasso ; Bayesian elastic-net ; Empirical Bayes ; Geometrically ergodic ; Gibbs samplers ; Importance sampling
Abstract:

In this paper we propose a novel geometrically ergodic Gibbs samplers for efficiently estimating the tuning parameters in Bayesian Lasso and Bayesian Elastic-Net. In the Bayesian framework, the shrinkage parameters can be estimated using either an empirical Bayes (EB) approach or a fully Bayesian analysis with an appropriate prior on the tuning parameters. There are several problems with the full Bayesian approach and also with EB method. In this paper we develop an EB approach with efficient importance sampling methods based on fast mixing Markov chains for estimating the shrinkage parameters in penalized regression methods.The other issue that we consider in this paper is the standard error estimation of the EN estimator. Our proposed methods for estimating the tuning parameters for Bayesian Lasso and Bayesian Elastic net is significantly faster than the traditionally used EM algorithm. We have also provided a full mathematical proof of the geometric convergence of our Gibbs samplers for Lasso and Elastic net. The practical effectiveness of our methods are illustrated by several simulation studies and two real life case studies.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2015 program





For program information, contact the JSM Registration Department or phone (888) 231-3473.

For Professional Development information, contact the Education Department.

The views expressed here are those of the individual authors and not necessarily those of the JSM sponsors, their officers, or their staff.

2015 JSM Online Program Home