Modeling Within-day Dynamics in Activity Scheduling: A Markov Decision Process Approach

抄録

This paper presents a modeling framework for the choice of activity type, timing and duration in activity scheduling. The traveler's choice reveals a strong dependency on the time of day and the activities that have been pursed before the current choice. The traveler also exhibits forward-looking behavior, i.e. the traveler realize the impact of the current choice on the future utility and take into account the future utility that he can obtain. The focus of this paper is thus capturing these dynamics in activity scheduling and replicating the traveler's daily activity plans. The activity scheduling behavior is formulated as a Markov Decision Process. Solution algorithms are developed for the model and numerical examples are presented.