A sequential dynamic choice model to simulate demand in evacuation conditions

When a dangerous event occurs, a variety of events affects the system characteristics of users and of the transportation network in the time. Dynamic models allow us to simulate variation in choice probability from one time to another, considering temporal evolution of user characteristics and of dangerous event. Among dynamic models, sequential dynamic discrete choice models represent a special class and are proposed in this work to simulate evacuation conditions. Sequential tests are introduced to validate the proposed model and in order to ascertain whether current decisions are directly influenced by the most recent previous decisions. Sequential tests are specified for evacuation condition simulation and allow us to assess the significance of the reduction in uncertainty