Using Big Data from TOTOR ETS to optimise public transport operations

Publication Type:
Thesis
Issue Date:
2020
Full metadata record
This thesis contributes to development of the trans-disciplinary field of Transport Analytics that aims to better target customer preferences and needs while optimising public transport operations. It demonstrates how use of empirical data acquired from Electronic Ticketing Systems (ETS) such as the Opal card system in Sydney, can provide more accurate and unexpected insights into demand patterns for public transport services. Systems that record every Tap-On and Tap-Off (TOTOR) pair, such as Opal, effectively provide a census of passenger responses to the services delivered, potentially increasing certainty and consensus on key aspects of operations such as required capacity, appropriate frequency and interchanging. Previously, the lack of detailed empirical data led public transport service providers to rely on top-down system-level models of macro-behaviour. The availability of high resolution TOTOR datasets provides an opportunity to develop bottom-up human-level models of micro-behaviour that can then be used to construct more accurate macroscopic system models. As will be shown, this difference in approach can lead to significantly different conclusions about patronage and appropriate service levels. The thesis approaches this new data and analytical opportunities in two ways. Firstly, by acknowledging and addressing concerns about privacy through development of a method to construct privacy-safe datasets; and secondly through new analytical methods to take advantage of the TOTOR data. The travel histories of individual customers contained within TOTOR datasets provide detailed biographical information; and so, to protect the privacy of individuals, access to these datasets has been highly restricted. In response, this thesis describes the methodological barrier created by the need for privacy protection, proposing a method to overcome this. The thesis provides several example case studies undertaken using analytics developed for activity datasets to improve public transport operations. These leverage inherent aspects of the data that were not available in previous data forms. The methods proposed leverages the ability to transform biographical-datasets into privacy-safe activity-datasets through deidentification, disassociation, aggregation and elimination. The method proposes three stages of transformation to create three levels of activity datasets with increasing privacy that can be distributed and shared appropriately between service providers and coordination agencies, to collaborators (such as researchers), and then to the general public. At all times, the research has been carried out within a customer-centric (customer service) approach to public transport service delivery. In the case studies, improved analytics has been shown to assist service partners in analysing and interpreting passenger behaviour in transport operations.
Please use this identifier to cite or link to this item: