大数据驱动的公共交通系统出行方式选择特性研究

掌握出行者选择偏好对于提高公共交通系统吸引力和优化运营策略有重要作用，且数据检测技术的发展使得交通大数据的收集更加便捷.本文以北京市出行大数据为依托，通过分析与建模，挖掘人们在公共交通中的方式选择特性，并探讨如何为交通运营提供决策支持.首先，对原始数据进行预处理、数据清洗与数据集成，获得了模型所需的数据集.然后根据效用函数理论，建立了分层多项Logit模型，并对模型进行求解与验证.最后，基于模型结果得出不同出行群体VOT，进行部分变量的敏感性分析.得出中距离出行者票价敏感性最高，短途出行者速度敏感性最高，6:00-7:00的出行群体票价敏感性最高且速度敏感性最低等结论，可用于支持公共交通运营部门制定最佳运营策略.

Travel Mode Choice Modeling and Analysis for Public Transportation System: A Big Data-driven Approach

Prioritizing public transportation development is highly promoted by transportation agencies worldwide to mitigate urban traffic common problems. Given the rapid development of traffic data collection technology, it provides an alternative way to mine people’s traveling preference using multi-source traffic big data. In this work, we firstly pre-processed, cleaned and integrated the data. With these data, a Multi- nominal Logit (MNL) model is built up and calibrated to describe travelers’discrete choice behavior. Then we verify the results in the light of Supervised Learning method. Finally, some insightful findings reveal the traveler’s sensitivity to price and speed that can support public transportation policy making. Compared to the traditional household travel survey, our research is demonstrated to be a promising alterative approach of transportation demand modeling.