Learning-Based Departure Time and Path Choice Modeling for Transit Assignment Under Information Provision: Theoretical Framework

摘要: The modeling of service dynamics has been the focus recent developments in field transit assignment modeling. This paper presents theoretical development a departure time and path choice model based on Markovian Decision Process. is core MILATRAS - MIcrosimulation Learning-based Approach to TRansit ASsigmnet. Passengers, while traveling, move different locations network at points (e.g. stop, board), representing stochastic process. process partly dependent performance controlled by rider. can be analyzed as In an MDP, actions are rewarded hence passengers? optimal policies estimated. proposed learning-based considers choice, stop run (or sequence runs) choice. classified bounded rational model, with constant utility term rule. appropriate for information provision since it distinguishes between individual?s experience provided about system dynamic characteristics.