ABSTRACT
Both the amount and number of sources of data is rapidly increasing in our society and thus also in the public transport industry. Almost all buses, trams and metros in the world are equipped with on-board computers and transmit terabytes of data with regard to for instance trip times, delays
and dwell times. Imagine social media data, such as user data of Twitter, Facebook and Flickr, which may yield new insights on public transport usage (Bregman 2012). Furthermore, video cameras (e.g. surveillance systems in stations and on-board vehicles), Wi-Fi and Bluetooth trackers provide information of pedestrian flows in stations, at platforms and in-vehicles (Van den Heuvel et al. 2015). Sensors connected to different types of assets, signals and switches for instances, enable optimization of maintenance schemes. This chapter focuses on a potential application of smart card data. Recently, many cities and regions introduced a smart card system for their public transport systems as discussed in this book and various publications such as Pelletier et al. (2011), Ma et al. (2013), Kurauchi et al. (2014), Wang et al. (2011) and Park et al. (2008). In addition to ticket handling, being an alternative for individual regional or urban tickets, these systems also provide valuable data. Without these systems, detailed information of origin and destination, number of passengers, trip lengths, etc. can only be collected by time consuming and expensive surveys. That is why the current surveys provide limited data sets. Smart card systems have the potential of providing more and better insights of passenger behaviour. These insights are helpful when dealing with the main challenges in the public transport industry. Within the public transport industry, we see several challenges. Due to the increased focus on cost savings, more attention to measures that increase cost efficiency of public transport is being paid. Meanwhile, passengers require higher quality of services. Although both developments seem to contradict each other, measures that serve both objectives do exist. Improving operational speed and service reliability, for instance, will lead to higher quality and lower costs at the same time, as shown by Van Oort et al. (2015b). However, to find and optimize cost-effective measures, detailed data on performance and ridership. Fortunately, the amount of data is increasing rapidly. Automated Vehicle Location (AVL) data has already been available for a long time (Furth et al. 2006, Hickman 2004) and recently much more passenger data (Automated Passenger Counting (APC) data) has become available as well (Pelletier et al. 2011). These data support public transport design and decision making, since they enable planners to illustrate the costs and benefits of certain problems and their solutions, for instance the added value of holding (Cats et al. 2012 and Van Oort et al. 2012) or optimized synchronization between tram and train (Lee et al. 2014). These costs and benefits are relevant for decision making and may be incorporated in cost-benefit analyses. This chapter deals with the Dutch smart card system, the so-called OVChipkaart and illustrates a potential application of its data. Our objective is to process the data in such a way that it enhances evaluation and prediction of ridership (patterns). This helps to improve network and timetable design. The main contribution of this research is to introduce smart card as a data source into existing methods to come to a new ridership
prediction approach. The outline of this chapter is as follows. Section 2 will elaborate on smart card systems in general and the Dutch smart card data specifically. Section 3 introduces our methodology to predict ridership. Section 4 is a case study, which reviews the applicability of our approach to data to predict future ridership. The conclusion and reflection on the approach are provided in Section 5. The methodology and case studies in this chapter are partly based on Van Oort et al. (2015a).
