Mining Public Transport User Behaviour from Smart Card Data

In urban public transport, smart card data is made of millions of observations of users boarding vehicles over the network across several days. The issue addresses whether data mining techniques can be used to study user behaviour from these observations. This must be done with the help of transportation planning knowledge. Hence, this paper presents a common “transportation planning/data mining” methodology for user behaviour analysis. Experiments were conducted on data from a Canadian transit authority. This experience demonstrates that a combination of planning knowledge and data mining tool allows producing travel behaviours indicators, mainly regarding regularity and daily patterns, from data issued from operational and management system. Results show that the public transport users of this study can rapidly be divided in four major behavioural groups, whatever type of ticket they use. Copyright © 2006 IFAC.