A Survey on Activity-Travel Pattern Reconstruction: Data Collection and Mathematical Models

Activity-travel patterns provide a behavioral description of daily mobility and support travel-demand forecasting, dynamic origin-destination estimation, and activity-based simulation. With the growth of passively collected mobility data, large-scale reconstruction has become increasingly feasible. However, these data are often incomplete and lack activity semantics, making it difficult to directly obtain complete activity-travel chains from raw observations. This paper reviews activity-travel pattern reconstruction from the perspectives of data collection and mathematical modeling. It first defines the reconstruction task by linking partial mobility observations with latent activity-travel chains. It then discusses major mobility data sources and explains how different observation mechanisms affect model design. Existing methods are grouped into model-driven, data-driven, and hybrid approaches, and their assumptions, advantages, and limitations are compared. Evaluation methods are further summarized at the element, chain, and population levels. The review suggests that future studies should focus on complete-chain inference, uncertainty representation, model transferability, and standard benchmarks. This survey provides an integrated framework for understanding and advancing activity-travel pattern reconstruction.