A spatially aware method for mapping movement-based and place-based regions from spatial flow networks

Community detection (CD) is a frequent method for analysing flow networks in geography. It allows us to partition the network into a set of densely interconnected regions, called communities. We introduce a new technique for including geographical weighting in existing methods for detecting spatially coherent communities. We take a link-based CD algorithm and adjust it to incorporate geographical weighting. We call this approach geographically weighted community detection (GWCD). Our method is demonstrated on two case studies of commonly encountered flow networks: commuter flows and taxi pick-up/drop-off flows. Further, we test different measures of distance for geographic weighting and compare our results with the unmodified CD algorithm. Our results show that GWCD can capture the geographical nature of flow regions, generating spatially smaller and more compact areas than if geography is omitted, and that it can be used to distinguish between different types of movement-type communities.

Automated summary

Community detection is a common method for analyzing flow networks in geography, allowing the partitioning of networks into densely interconnected regions. Geographically weighted community detection can capture the geographical nature of flow regions, generating smaller and more compact areas, and can be used to distinguish between different types of movement-type communities.