Public transport trip purpose inference using smart card fare data

• Published in: Transportation research. Part C, Emerging technologies Emerging technologies, 2018-02, Vol.87, p.123-137
• Main Authors: Alsger, Azalden, Tavassoli, Ahmad, Mesbah, Mahmoud, Ferreira, Luis, Hickman, Mark
• Format: Article
• Language: English
• Subjects: O-D trips; Public transport smart card fare data; Rule-based modelling; Trip purpose
• ISSN: 0968-090X; 1879-2359
• DOI: 10.1016/j.trc.2017.12.016

•This paper proposes a method to infer passenger’s trip purpose from smart card fare data by integrating several attributes and databases.•The model validation results show a strong capability to predict trip purpose at a high level of accuracy.•Different trip purposes show different sensitivities to the applied spatial and temporal attributes.•The level of the available land use details is the main cause of errors.•The frequency of the trips shows as a potential attribute to confirm the inferred results. Although smart card fare data has recently become more prevalent as a rich, comprehensive and continuous source of information, there is still some missing information which inhibits its capability in the research field. One key missing piece of information is the passengers’ trip purpose. This paper investigates the potential of the smart card data to infer passengers’ trip purpose, thereby reducing the use of the expensive and time-consuming Household Travel Surveys (HTS). On this basis, an improved model has been proposed, calibrated and validated for trip purpose inference by integrating different data sources, namely: HTS, a land use database, the South East Queensland Strategic Transport Model (SEQSTM), the General Transit Feed Specification (GTFS) data, O-D survey data, and most importantly the unique smart card fare data from Brisbane, Queensland. As smart card fare data does not record passengers’ trip purpose, the calibration and validation procedures are performed on HTS data. Based on the validation results, the proposed methodology shows a strong capability to predict trip purpose at a high level of accuracy. The results show an overall 67% correct inference after applying spatial attributes, but the correct inference increases to 78% after applying temporal attributes. Different trip purposes show different sensitivities to the applied spatial and temporal attributes. Work and home trips have the highest correct inference results, at 92% and 96%, respectively. Furthermore, the results of correct inference for shopping and education trips improved after applying the temporal attributes.