Travel Pattern Recognition using Smart Card Data in Public Transit

Y. Chang,He Zhao-cheng
2016-01-01
Abstract:As the basic travel service for urban transit, bus services carry the majority of urban passengers. A better understanding of transit riders’ travel characteristics can provide a first-hand reference for the evaluation, management and planning of urban public transport system. Over the past two decades, 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 for mining smart card data is developed to recognize the travel patterns of transit riders. A smart card dataset is first processed to obtain the trip information. After reconstructing the transit trip chains from the trip information, this paper adopts the densitybased spatial clustering of application with noise (DBSCAN) algorithm to mine the historical travel patterns of each transit riders. In addition, a sensitivity analysis is conducted to evaluate the optimum parameters. In case study, the analysis of travel pattern characteristics is conducted focusing on the transit riders of Guangzhou City, China.
What problem does this paper attempt to address?