Urban Area Division and Function Discovery Based on Trajectory Data

Abstract

Urban functional area identification can not only help city managers to carry out urban planning effectively, but also have important significance in policy making and resource allocation. In this paper, we propose a method of urban functional area identification based on data mining of human large-scale spatiotemporal trajectory and the semantic analysis of POIs, and construct the analytic model UADAFIM (urban area Division and Function identification model). Adaptive density clustering through trajectory data, by clustering each class cluster and producing the central point of each cluster, we use the central point as the discrete points to produce the Voronoi Diagram, and through our algorithm we divide the city into different areas, effectively avoiding the restriction of dividing city by the urban road network. In addition, we make the regional functional theme documents by the POIs data, the LDA model is used to effectively excavate the main functions of each area. Our study takes the Beijing city as the analysis, the research object, as well as the verification object.