Uncovering Spatio-temporal cluster patterns using massive floating car data

Abstract

In this paper, we explore spatio-temporal clusters using massive floating car data from a complex network perspective. We analyzed over 85 million taxicab GPS points (floating car data) collected in Wuhan, Hubei, China. Low-speed and stop points were selected to generate spatio-temporal clusters, which indicated the typical stop-and-go movement pattern in real-world traffic congestion. We found that the sizes of spatio-temporal clusters exhibited a power law distribution. This implies the presence of a scaling property; i.e., they can be naturally divided into a strong hierarchical structure: long time-duration ones (a low percentage) whose values lie above the mean value and short ones (a high percentage) whose values lie below. The spatio-temporal clusters at different levels represented the degree of traffic congestions, for example the higher the level, the worse the traffic congestions. Moreover, the distribution of traffic congestions varied spatio-temporally and demonstrated a multinuclear structure in urban road networks, which suggested there is a correlation to the corresponding internal mobile regularities of an urban system.