摘要
公共交通出行者的出行特征是地铁及公交线网规划与运营优化的重要依据。基于多模式刷卡数据,提出城市公共交通系统出行链提取方法,利用存在换乘的出行链调查数据进行验证,提取成功率达 96.1%。基于出行者历史刷卡数据构建了多种机器学习分类器以识别通勤人群,经过精度比较,发现随机森林分类器效果最优,准确度达 99.96%。利用分类器和出行链提取方法,对北京市公共交通系统出行链结构、换乘特征等进行初步分析。该方法可以有效提取分析通勤人群出行特征,为公共交通系统方案的优化提供数据支持。
The trip characteristic by public transit is an important basis of public transit network planning and operation optimization. Relying on the multiple-mode smart card data,an extraction method of public transit trip chain(PTTC) is proposed,and the extraction success rate is 96.1%. Classifiers of machine learning are formed to distinguish the commuters and non-commuters based on the history card data of passengers. Through accuracy comparison, the Random Forest model classifier presents the optimal effect with 99.96% accuracy. The structure of PTTC and the transfer characteristics in Beijing are analyzed by using the proposed classifier and trip chain extraction method,which can effectively extract and analyze the characteristics of commuters and provide support for the optimization of public transit system.
作者
翁剑成
涂强
袁荣亮
王月玥
WENG Jiancheng;TU Qiang;YUAN Rongliang;WANG Yueyue(Beijing Key Laboratory of Traffic Engineering,Beijing University of Technology,100124,Beijing,China)
出处
《城市轨道交通研究》
北大核心
2019年第6期66-71,共6页
Urban Mass Transit
基金
国家自然科学基金(51578028)
交通运输部科技计划(2015318221020)
关键词
公共交通
地铁
出行特征
出行链
机器学习
public transit
metro
trip characteristic
PTTC
machine learning
作者简介
第一作者:翁剑成,副教授.
