Regression-based monitors of North American border-crossing activity

Abstract

Official border-crossing data were collected, assumed to represent usual activity and summarised with linear regression models to facilitate residual control charts and detect unusual activity that might be attributed to a security threat or some other external variable. Results are presented for northern highway and loaded rail, containers at North American borders with respect to the sigma limits of statistical quality control. Additionally this article describes the estimation of what may be useful model-based charts for activity across three ports of the State of Michigan: Detroit, Huron and Sault Ste. Marie.