ABSTRACT

Understanding the travel behavior of individuals grouped by similar time-use activity patterns can contribute greatly to modeling regional spatial and temporal patterns of transport demand. In this paper we present a comprehensive modeling framework to forecast and replicate individuals’ travel behavior, labeled as the Scheduler for Activities, Locations, and Travel (SALT). The prototype version of the SALT framework comprises a series of modules that employ behaviorally-based econometric, machine-learning, and data-mining techniques. The SALT model is cross-validated with 30% of the out-of-home sample survey data from the large Halifax Space Time Activity Research (STAR) household survey. Results show that the SALT scheduling model is able to assemble the travelers’ 24-hour schedules with an average 82% accuracy compared to the observed data. The proposed simulation modeling framework is useful for deeper understanding of individuals’ activity-travel decisions and may be utilized to examine sensitive policy issues such as transportation control measures and congestion-pricing.

摘要

了解按相似的时间使用活动模式分组的个人的旅行行为，可以极大地有助于对运输需求的区域时空模式进行建模。在本文中，我们提供了一个全面的建模框架来预测和复制个人的旅行行为，称为活动，位置和旅行计划表（SALT）。SALT框架的原型版本包括一系列模块，这些模块采用基于行为的计量经济学，机器学习和数据挖掘技术。SALT模型与来自大型Halifax时空活动研究（STAR）家庭调查的30％的户外样本调查数据进行了交叉验证。结果表明，与观测数据相比，SALT计划模型能够以平均82％的准确度来组合旅行者的24小时计划。