OREANDA-NEWS. Deutsche Bahn is about to begin using a new data analytics service provided by T-Systems to improve their existing real-time system for projecting arrival and departure times on their rail passenger services. The timetable data for more than two million stops per day for essentially the whole of Deutsche Bahn's scheduled passenger services will be compared minute-by-minute against the current real-world transport situation. This comparison is then used as the basis for forecasting probable arrival times and, at the same time, for predicting the impact of the real-time information on subsequent passenger transfer connections.

From T-Systems' data centers, the system analyzes geo-positioning reports received for all trains currently running in a matter of seconds, and produces a projection of expected arrival times up to the trains' final destinations. The algorithm used in these calculations is based on machine learning (artificial intelligence). To achieve its results, the algorithm chooses from a variety of projection models depending on the current traffic situation. At 24-hour intervals, the model trains itself up on the basis of historical data during the night. This self-training process helps to improve the accuracy of the system's prognoses continuously and to adapt them to current conditions on the transport network. The prognostic system is based on a solution developed in-house by T-Systems and its subsidiary T-Systems Multimedia Solutions, and will be further developed and implemented within a joint project with Deutsche Bahn.