An Extended Community Detection Algorithm to Compare Human Mobility Flow Based on Urban Polycentric Cluster Boundaries: A Case Study of Shenzhen City

Urban polycentric spatial structure is expected to play an important role in organizing city activities and economies in urban planning. Polycentric structure is a very popular form to organize urban space especially for new cities to attract human activities within each cluster area and reduce long distance travel between locations. In principle, each cluster area forms a human mobility community and services that population. However, few studies have examined differences between the nominal human mobility community boundary and the planned urban polycentric cluster boundary. This paper proposes an extended community detection algorithm based on CNM community detection algorithm, introducing source nodes, which generate communities by integrating the initialized source partitions. Using this extended algorithm, this paper compares human mobility community boundaries and planned urban polycentric cluster boundaries for the case of Shenzhen City. Most polycentric clusters match the human mobility community well, which indicates that they successfully attract activities. There were a few polycentric clusters with large differences from the detected community, which indicates that these community areas were not executed and/or planned well, and more attention should be given in future cluster design when dividing service areas.