Project Summary:
With the growing academic focus on mobility behaviors in human geography and transportation studies, this project investigates multi-scale mobility patterns among Chinese urban residents, including residential relocation, job mobility, and daily commuting. Adopting a holistic trajectory-based approach, it emphasizes the sequential decision-making processes and multi-scale interactions inherent in mobility behaviors. To address the multidimensional complexity of trajectory data and dynamic decision sequences, the study first constructs an analytical framework integrating machine learning (ML) algorithms for mobility trajectory pattern mining and sequential decision simulation. Supervised and unsupervised learning techniques are employed to identify routine and anomalous behaviors, while inverse reinforcement learning models simulate decision-making sequences. Second, the research unravels the mechanisms underlying mobility decisions, including the role of behavioral habits, path dependency, prospective/lagged effects, dynamic interactions with external environments, and cross-temporal linkages between short- and long-term decisions. Finally, empirical analyses reveal residents’ evolving preferences for social environments and built environments across mobility scales, offering dynamic assessments to guide spatial optimization strategies.

Principal Investigator: Zhang Wenjia
Duration: January 2019 – December 2021