In October, I attended the Association of Collegiate Schools of Planning (ACSP) Annual Conference held in Minneapolis, Minnesota. This year’s conference brought together planning educators, researchers, and practitioners from around the world to exchange ideas on advancing resilient and equitable urban planning, governance, and education.
Presenting on National-Level Commuting Flow Modeling
I presented my paper, “Enhancing Home–Work Commuting Prediction for Decarbonized and Climate-Resilient Transport Planning” during the conference session: ‘Complexity Science and Urban Systems’.
My presentation introduced our ongoing work on graph neural network (GNN)–based approaches for modeling commuting patterns across the UK, which integrates socioeconomic data and transport accessibility indicators. The goal is to improve projections of commuting flows under future development and policy scenarios, supporting transport planning that aligns with decarbonization targets and climate adaptation goals. Our session generated insightful discussions on the potential of machine learning and network-based modeling to capture mobility dynamics and inform more sustainable transport decision-making.
More broadly, the work contributes to the DARe research agenda by highlighting how national-scale mobility modeling can help evaluate low-carbon transition pathways, identify vulnerabilities in transport networks, and further support activity-based models to assess impacts across regions and social groups. Several questions from the audience focused on how such models could support scenario-based planning, and the challenges of epistemic uncertainty in deep learning models.
