Region-Wise Variable Selection with Bayesian Group Lasso

• Activity Number: 663
• Type: Contributed
• Date/Time: Thursday, August 4, 2016 : 8:30 AM to 10:20 AM
• Sponsor: Business and Economic Statistics Section
• Author(s): Sayan Chakraborty* and Tapabrata Maiti
• Companies: Michigan State University and Michigan State University
• Keywords: Group Lasso ; Spike and Slab Prior ; Conditional Autoregressive Structure ; Median Thresholding ; Bessel Function

Abstract

A common spatial variable selection problem in these days is to select the variable and the corresponding coefficient estimates for different locations. We investigate this problem using a bayesian approach by introducing Bayesian Group LASSO technique with a bi-level selection which not only selects the relevant groups but also selects the relevant variables within group. We use spike and slab prior along with the Conditional Autoregressive Structure among the model coefficients which validates the spatial interaction among the covariates. Median thresholding is used instead of posterior mean to have exact zero's for the variables which are not important. We untimately perform simulations to show that the method discussed in this paper does an excellent job in selecting as well as estimating the relevant variables.