Comparison of Agent-Based Modeling and Equation-Based Modeling for Transportation Behavioral Studies

摘要

Nowadays there are two main approaches to simulating traveler's behavior: equation-based modeling (EBM) and agent-based modeling (ABM). In equation-based modeling, the most commonly used model is Logit model, which uses a utility function to aggregate and evaluate variables of system and entities. While in agent-based modeling, a set of production rules are developed to relate a traveler's characteristics to the choice he will make. Each approach has its advantages and disadvantages, which are determined by its assumptions and implementation methods. This paper compares these two models thoroughly from the perspective of theory and assumptions. Then we conducted a joint revealed/stated-preference paper-based and web-based travel behavior data survey in Beijing. Based on the survey data, we built an agent-based model and an equation-based model to catch the travel behavior switch under different conditions. Then we compared the prediction accuracy, data dependence and model stability between the two models. The results indicate that prediction accuracy of both models is on similar level. Though ABM depends more on data size, it is more stable under noisy data. According to the characteristics of the two models, a set of criteria of modeling selecting is developed between ABM and EBM under various circumstances.