Understanding Human Mobility Within Metro Networks – Flow Distribution and Community Detection

In this paper, smart card data collected from the Nanjing Metro over 2-hour time periods are used to characterize within- and between-day human mobility patterns within the metro network. Results show that the OD (origin to destination) flows can be characterized well by shifted power law distributions with similar exponents around 2, which reflects the fact that a few OD pairs in the system play a dominant role and undertake disproportionately large OD flow distribution. The different exponents signify heterogeneous human movement in within- and between-day ranges. In addition, we analyze the metro community structures over different time periods based on the community detection method using random walks to visualize and understand passenger movement from a spatial perspective. Normalized mutual information is used to compare community partitions over different time-intervals. The results show that the properties of human mobility during different time periods have a similar rhythm, although some nuances exist, and the community structure is usually divided according to the line distribution. This empirical study provides spatiotemporal insights into understanding urban human mobility and some potential applications for transportation management