基于离散-连续选择模型的通勤出行时间预测

Prediction of commuter's time allocation with discrete-continuous choice model

把握出行者的日时间分配是交通行为分析的重要内容之一.本文基于离散-连续建模思想,结合运用Ordered Probit离散选择模型和Hazard连续选择模型,建立了由上班(上学),下班(放学)出发时刻模型和上班(上学),下班(放学)出行耗时模型组成的通勤时间预测模型系统,预测了通勤者的日时间安排.研究表明,所建模型能够以较高的预测精度,预测通勤者的活动-出行时间安排.研究将为活动-出行行为的整体建模预测和分析奠定时间轴预测基础,为城市居民的交通行为分析提供模型工具,为制定交通管理政策,解决城市交通拥挤问题提供决策分析依据.

Time allocation is an important issue in travel behavior analysis. Based on the theory of discrete-continuous modeling, this paper presents a model system for commuter＇s daily time allocation, which is composed of two models for departure time forecasting and two for travel duration prediction. Ordered probit model and hazard model are employed for discrete and continuous modeling, respectively. Based on the model system, commuter＇s daily time allocation is predicted. The results indicate that the goodness-of-fit of the model system is acceptable. This study can contribute to the development of a comprehensive full-scale model of daily activity-travel patterns. It also provides basis for the theoretical work about travel behavior analysis and modeling as well as the empirical work of transportation planning and management.