Robust Adaptive Group WLAD-LASSO

• Activity Number: 309
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Nedret Billor* and Kristin Lilly
• Companies: Auburn University and Auburn University
• Keywords: Group LASSO ; LAD-LASSO ; Oracle Property ; Variable selection

Abstract

In this study, we propose four group variable selection methods, the group LAD-LASSO and the group WLAD-LASSO and their adaptive counterparts, to perform variable selection at the group level. These methods are designed to excel at selecting important groups when the data include outliers. Some theoretical details are shown, including the existence of the oracle property. A simulation study and application to a real data set is presented for all methods.