Perspectives on Profiling

• Activity Number: 186
• Type: Topic Contributed
• Date/Time: Tuesday, August 10, 2004 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Defense and National Security
• Author(s): Jeff Butler*+ and Steven Smith*+ and Samuel E. Buttrey*+
• Companies: Bureau of Transportation Statistics and U.S. Department of Justice and Naval Postgraduate School
• Keywords: profiling ; clustering ; classification trees ; fraud detection ; data-mining

Abstract

"Profiling'' individuals according to characteristic traits suggests infringement on civil liberties. This session will consider three different perspectives on profiling: Jeff Butler on transportation, Steven Smith on legal aspects: "Statistical and Legal Issues in Profiling and Related Activites." This presentation will address legal and data collection issues and experiences related to generating estimates on racial profiling, primarily in law enforcement. And Sam Buttrey on classification: Clustering techniques divide observations into groups but often depend on the scaling of the variables, are changed by monotonic transformations, and do not provide for selection of "important" variables. We fit a set of regression or classification trees with each variable acting in turn as the "response" variable. Trees with poor predictive power are discarded. Therefore, "noise" variables will often appear in none of the trees and have no effect on the clustering. The technique is unaffected by linear transformations of the continuous variables and is resistant to monotonic ones. Categorical variables are included automatically.