Statistical Designation of Traffic Control Subareas

Abstract

A method for identifying control subareas in traffic signal networks has been developed and tested with data from the Miami, Florida, business district. The procedure combines the use of cluster and discriminant analysis to group signalized intersections. Unlike existing procedures which use degrees of association between adjacent signals, the multivariate procedures used in this method allow for consideration of degrees of association among all signals in the original network. Model tests used the flow ratio for each intersection approach as the vector of intersection attributes. The resulting subareas were compact, well defined and slightly different for morning and midday traffic conditions. The procedure appears to have considerable utility in defining control subgroups in signal system design.