考虑出行模式和周期性的公交出行特征分析

Analyzing Methods of Residents' Travel Characteristics Considering Travel Patterns and Periodicity

公交服务承载着广大城市居民日常的出行需求,深入了解居民公交出行特征是公交管理部门的迫切需要.近年来,IC卡数据已成为公交出行调查数据的新来源,能提供更全面的时空信息.本文考虑居民公交出行模式及周期性特征,提出了一套居民出行特征分析框架.首先基于IC卡数据的挖掘获取居民个体出行信息,应用DBSCAN聚类算法来检测居民的历史出行模式,并提出了基于模式划分的出行周期性分析方法;最后结合多项出行特征指标,利用K-means++算法对居民出行规律性进行聚类评价.在实例分析中,以广州市居民作为研究对象,开展了居民出行特征分析.实验结果表明了该方法的有效性.通过本文方法挖掘的信息对提升公交服务政策具有重要意义.

Public transit services carry the majority travel demand of urban residents. There is an urgent need for transit authorities to make a better understanding of the travel characteristics of transit riders.Recently, data from smart cards have become a new source of travel survey data, providing more comprehensive spatial-temporal information about urban public transport trips. In this paper, a methodology considering travel patterns and periodicity is developed to analyze the travel characteristics of residents. The individual trip information is firstly obtained based on smart card data and transit riders＇ trip chains are identified. The DBSCAN clustering algorithm is applied to examine transit riders＇ historical travel patterns.And a methodology is then proposed for analyzing the travel periodicity based on travel patterns. Finally Kmeans ＋＋ algorithm is applied to cluster and evaluate the travel regularities of transit riders. In case study,focused on the transit riders of Guangzhou City, the analysis of travel characteristics are conducted. The results proved the effectiveness of the proposed method. The information mined through the method is of great significant for improving transit service strategies.