融合多源数据的城市公共交通需求预测方法

发布时间:2024-04-10        浏览量:10

时间:2024年4月11日(星期四)13:30-14:30

地点:经管大楼A楼 四楼第二会议室报告厅

主题:融合多源数据的城市公共交通需求预测方法(The Travel Demand Forecasting of Urban Public Transit System by Fusing Data from Multiple Sources)

主讲人:王可(上海理工大学管理学院)

简介:王可,信管系讲师,硕士生导师,同济大学博士,美国德州大学奥斯汀分校公派联合培养博士。主要研究方向为融合人工智能的交通需求建模、出行行为分析、大数据挖掘等。主持上海市哲学社会科学规划课题,参与国自然重点项目、面上项目、上海市社发科技攻关等项目。在TransportationTRRSCI/SSCI检索期刊发表论文10余篇,美国科学院交通研究委员会论文10余篇。

WANG Ke, Lecture of Department of Information Management and Information Systems, Supervisor of Master graduate student, Ph.D. of Tongji University, Joint Ph.D. student of the University of Texas at Austin. Research field: travel demand modeling by integrating artificial intelligence, travel behavior analysis, and data mining. PI of a project from the Shanghai Planning Office of Philosophy and Social Science. Participate in several projects, such as key and general projects from the National Natural Science Function of China and the Science and Technology Commission of Shanghai Municipality. Published more than 10 academic papers in Transportation and Transportation Research Record, and more than 10 papers in the Annual Meeting of Transportation Research Board of American National Academies of Sciences.

摘要:融合多源数据分析城市交通出行方式选择的影响因素,并预测关键因素对于公共交通出行方式市场份额的影响程度。探索混合多项Probit模型随机系数的可辨认性,构建基于ProbitSP-off-RP模型,构建Gumbel分布及其有限混合产生服从MEV分布的随机数。采用多元正态分布的解析逼近算法,克服以往仿真积分和有限极大似然估计等传统估计方法的缺陷。结合上海市居民通勤方式选择的因素,并基于弹性分析进行公共交通的需求预测,展示基于个体的交通需求模型在城市规模上的应用价值。 

Integrate multi-source data to analyze the influencing factors of urban transportation mode choice, and predict the impact of key factors on the market share of public transportation mode.Explore the identifiability of random coefficients of the mixed multinomial Probit model, develop the Probit-based SP-off-RP model, and use the Gumbel distribution and its finite mixture to generate random variables that follow the MEV distribution. The analytical approximation algorithm of multivariate normal distribution is used to overcome the shortcomings of traditional estimation methods such as simulation integration and limited maximum likelihood estimation.Combining Shanghai residents’ commuting mode choice data and forecasting public transportation demand based on elasticity analysis, the application value of the individual-based transportation demand model at the urban scale is demonstrated.