A Penalized Regression Approach in Detection of Nuclear Materials in Shipment to the United States

• Activity Number: 311
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Defense and National Security
• Author(s): Xueying Chen* and Jerry Cheng and Minge Xie+
• Companies: Rutgers University and Rutgers University and Rutgers University
• Keywords: Penalized Regression ; Correlation Analysis ; Two-level Selection ; Application to Homeland Security

Abstract

This paper proposes a penalized regression method based on a penalty function associated with the correlation structures of explanatory variables in addition to the usual penalty on coefficients. This method not only incorporates the correlation patterns leading to the underlying group structure, but also performs variable selection at both between-group and within-group levels. It is shown that the proposed method has a general grouping effect and achieves sparsity property in the sense of removing unimportant groups. In the project motivated by a project sponsored by Department of Homeland Security, we analyze the information of shipping containers in various customs forms and develop models to assign risk scores for future shipments. The noisy data are analyzed using the proposed penalized regression method which facilitates detection of suspicious cargos and reduces the false alarms.