Abstract

Passengers can pay with an electronic smart card in every tram operational in Amsterdam. Malfunction of smart card validators leads to reduced customer satisfaction, loss of revenues and unplanned costs. Malfunction of a smart card validator is relatively rare: it fails about once every two years, which is recorded by the maintenance shop inaSAP database. The validators generate transactions and events. The events (about 30 a day per validator) are stored in the OV Chipcard database. During this project, weinvestigated whether it was possible to understand malfunctions by analyzingtheeventdatageneratedinthe24hourspreceding those registered malfunctions. Analysis was done on over six million events generated within a four-month period (January - April 2017) by more than 1,700 validators mounted in 200 trams. The selected decision tree models showed that about 50% of registered malfunctions were related to specific events that occur in relatively high frequencies. These events signified loss of communication and/or the inability to receive GPS location information. The use of decisiontreemodelsmadeitpossiblenotonlyto predict the malfunctions but also to get a better understanding of the root cause of the malfunctions. These insights can be used asinputforimprovingthereliabilityof the smart card validators.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call