Taxi Trajectory Clustering Based on Network Clustering Method

To effectively shorten the taxi trajectory sequence and remove redundant trajectory point of interest information, in practical applications, a taxi travel network is often constructed based on complex network modeling theory to explore the user’s daily travel rules and characteristics. This paper takes the taxi GPS trajectory data in Beijing as the research object, constructs a spatially embedded complex network that takes into account temporal factors, and analyzes the small-world attributes of it. Finally, different from previous methods of directly clustering trajectory data, this paper conducts a Louvain network clustering algorithm to divide the constructed travel network. Corresponding with the distribution of point of interests of interest in different communities, the division result taking into account the temporal factor will be used as another important data source for analyzing spatial-temporal characteristics of urban travel. This method has good usability in different cities and regions.