Analysis of Sparse Roadway Trajectories

Recent technological advances enable the gathering of extensive data on vehicular trajectories of large numbers of travelers at an unprecedented level of detail. Such trajectory datasets provide a wealth of information for purposes such as urban planning, carpool formation, and public-transportation design. This paper describes methods for analyzing and visualizing such data with an emphasis on sparse-traffic environments. It outlines the needs of applications in this domain and presents methods for clustering trajectories and for visualizing the results. The methods are evaluated by an experimental study on a publicly available dataset from real travelers.