Regional Delineation Based on A Modularity Maximization Approach

Abstract

Regional delineation is critical to urban policy formulation and infrastructure construction. For the convenience of regional management, the population flow in the same region should be as dense as possible, and that between different regions should be as little as possible. We consider the population flow as a kind of correlation between urban plots, and construct graph by using unit plots as nodes and population flow as the edges. By combining strategies of hierarchical aggregation and node movement, a novel community detection algorithm based on Modularity maximization is proposed. The efficacy of the proposed algorithm on Modularity optimization is verified through experiments using real world data set. Our method outperforms baselines on objective optimization with an acceptable execution time. Moreover, a case study in Nanjing China is presented, and the result demonstrates the rationality of the regional delineation from our proposal.