Javier Argota Sánchez-Vaquerizo, Dirk Helbing

Cities Volume 162, July 2025, 105901

Origin-destination (OD) matrices are essential for the analysis, planning, and simulation of urban areas, infrastructure, and transportation systems. However, they are often costly and time-consuming to determine, which reduces their potential use for informed decision-making and planning in cities. This research introduces a novel spatial econometric method that considers spatial spillover effects of socio-demographic, land use, and topological variables to directly estimate traffic OD flows between zones of the Metropolitan Area of Barcelona. Employing a two-part Hurdle model with gradient boosting (XGBoost), our approach achieves low error rates (MAE = 6.109, RMSE = 98.774), comparable to other established models also analyzed, but the proposed method’s simplicity facilitates its practical application in urban planning and policy-making. This is illustrated by applying the proposed model to predict changes in vehicle flows resulting from the conversion of offices into other urban uses such as housing, commerce, education, or storage. Despite the related population increase, we expect a reduction in vehicle trips by up to 10 % even with limited spatial interventions. Our findings suggest the model’s power to assess urban trends and policies, particularly in considering teleworking expansion, housing shortages, and contemporary planning practices promoting alternative mobility modes and densification. This research underscores the dual benefits of methodological innovation and practical policy application, marking a significant advancement in urban planning.

Read the full article at: www.sciencedirect.com

• Papers