| | | The objective of this project is to better understand activity sequences in the Panel L�manique GPS tracking dataset. In activity-based models, people do not travel for the sake of traveling, but because they need to perform activities such as working, studying, shopping, eating, accompanying someone, or participating in leisure activities. Understanding mobility therefore requires understanding how activities are organized in time and space, and how these patterns vary across individuals and across days.
Students will work directly with GPS-based mobility data. The first goal is to examine how activities were inferred from raw trajectories, assess the consistency of the detected activities, and identify possible limitations or sources of error. The second goal is to model activity sequences, from the representation of an average day to a richer multi-day perspective. The project combines statistical modeling, in particular discrete choice models, with optimization methods, since the simulation of possible activity-travel alternatives requires generating and evaluating feasible schedules. |