Segmentation and evolution of urban areas in Beijing: A view from mobility data of massive individuals

With the rapid urbanization of Beijing in recent decades, comprehensively understanding its regions' structures and functions becomes more and more challenging, though it indeed plays a fundamental role in the city planning. While fortunately, the accumulation of huge mobility records from massive individuals provides an unprecedented big-data window for solving this issue. In this paper, we segment urban areas of Beijing into administrative and functional subdivisions through mining GPS trajectories of taxis. First, a flow network between small regions is established to administratively segment the urban area and Infomap is found to be a better approach. Second, temporal features from regions' flow dynamics are extracted to functionally segment the urban area through spectral clustering, which effectively identifies regions with different functions and flow patterns. Third, the comparison of segmentation at different time can vividly represent the evolution of the city, including emergence of new regions and vanishment of aging areas. Our results demonstrate the possibility that the big-data of movements generated by massive users could provide a new but promising probe to understand the evolution of cities in both spatial and temporal dimensions.