The problem relates to the current way of routing train conductors/stewards through the railway system. During our internships we experienced this to be inefficient, which leads to a non-optimal quality of service and safety in trains, and a high percentage of fare-dodgers, whereas a data-driven routing optimization could improve on these quantifiers. Moreover, current systems are used in an offline way that are not robust to train delays or unexpected train personnel tasks. Furthermore, key performance indicators (KPIs) for train conductors, like checking tickets of 10% of the daily travellers or being present at all stations six times a day, can be hard to meet and hard to measure.
We propose an online data-driven optimization tool that routes stewards through the working day. We use the data on fines, passenger numbers, and incidents, for our optimization tool to send stewards to the “hot times'' and “hot spots” of incidents and fare-dodgers. This tool can reroute stewards in real-time, in case unpredicted events occur. Moreover, we propose an offline optimization tool to consider the costs and benefits of the number of stewards/train conductors against the quality of service and safety, taking into account applicable KPIs.