🚀 New paper out in Transportation Research Part B: Methodological - ISTTT26 Special Issue 📄 "Capturing Behavioral Heterogeneity for Traffic Flow: A Scalable and Personalized Decomposition Approach" Agent heterogeneity has long been a defining characteristic of traffic flow systems, yet it remains one of the hardest to understand and to model. We show in this work how approaches of learning traffic models under heterogeneous data can mislead inference. We then propose a modeling pipeline that gives each agent its own behavioral signature while preserving population-level structure. The framework opens doors to behavioral transfer across agents (think personalizing #AutonomousVehicles behavior using learned human driving styles), feature-targeted clustering, and richer interpretation of traffic dynamics from the bottom up. 🎤 We will be presenting this work as a lectern presentation at the International Symposium on Transportation and Traffic Theory (#ISTTT26) in Munich, Germany this July! 👥 Joint work with the amazing team: Yongju Kim, Xinzhi Zhong, and Soyoung Ahn 🔗 Read it here: sciencedirect.com/science/ar… #TransportationResearch #TrafficFlow #DriverBehavior #AutomatedVehicles #Personalization #Heterogeneity