Quantifying the spatial homogeneity of urban road networks via graph neural networks

Quantifying the topological similarities of different parts urban road networks enables us to understand growth patterns. Although conventional statistics provide useful information about characteristics either a single node’s direct neighbours or entire network, such metrics fail measure subnetworks capture local, indirect neighbourhood relationships. Here we propose graph-based machine learning method quantify spatial homogeneity subnetworks. We apply 11,790 across 30 cities worldwide within each city and cities. find that intracity is highly associated with socioeconomic status indicators as gross domestic product population growth. Moreover, intercity values obtained by transferring model reveal similarity network structures originating in Europe, passed on United States Asia. The development revealed using our can be leveraged transfer insights between It also address policy challenges including planning rapidly urbanizing areas regional inequality. quantified fine-grained manner graph neural networks. This studied inner-city around world used study help planning.