Mining the Temporal-Spatial Patterns of Urban Traffic Demands Based on Taxi Mobility Data

Urban traffic is a complex temporal-spatial process. Understanding the dynamical behavior of the whole urban traffic system will allow traffic organizers to identify the source of traffic congestion. In this study, we conducted an in-depth analysis of a taxi trajectory dataset in Beijing based on a dynamical graph and adopted a traffic-modified PageRank algorithm to evaluate urban traffic demands. By generating feature vectors, we have analyzed the temporal-spatial patterns of the distribution of traffic demands in Beijing. We obtained a general picture of the distribution of traffic demands in Beijing and also successfully extracted different zones with significant traffic demands. We discovered that most of Beijing's traffic demands lie on internal ring roads at daytime and on peripheral highways at nighttime, which suggests that the structure of road network and drivers' proneness for choosing quicker paths are still the most influential factors of urban traffic.