Deriving Truck Route Choice from Large GPS Data Streams

Abstract

This paper describes the development of a simple and effective procedure to derive truck route choice data from large streams of truck GPS data. At the heart of this procedure is a map-matching algorithm that is easy to implement. The proposed procedure is applicable to large data sets with varying frequencies between consecutive GPS records that range from once every second to once every 20 min. As a demonstration, the procedure is employed to derive routes for more than 78,000 truck trips from a large GPS data set. The derived routes are validated to demonstrate the accuracy of the results. The paper also introduces a useful metric for the measurement of the overlap between the travel path of one trip and that of another trip between a given origin–destination (O-D) pair. In addition, the paper discusses a procedure to employ the metric in the measurement of the diversity (or variability) of routes across several trips between a given O-D pair and to identify the set of unique routes used between the O-D pair.