Are transit users loyal? Revelations from a hazard model based on smart card data

Smart card fare collection systems for public transit produce a huge quantity of data on a daily basis. The ability to follow the use of a single card throughout the months gives the opportunity of measuring the loyalty of the individual to the service. Then, operators can have quantitative knowledge of the loyalty in their network. However, it is also important to know what are the factors that influence the survival of the users. This paper presents the application of a discrete time hazard model to 5 years of data of a medium-size transit authority in Canada. The concept of the hazard model relates to the fact that the probability to continue the use of a smart card by user i, at time t, is conditional to the probability of not cancelling the card before the time period. Hence, its use is appropriate in this case. Results for the regular adult fare show that loyalty is positively influenced by residential density and by the transit share in the area. A younger population will also be retained longer in the system. However, a high unemployment rate has a negative impact on survival. A high share of transit and walk trips is also affecting the loyalty, suggesting that the retention is reduced when there are more mode choices available to the commuter.