Partitioning of Public Transit Networks

Partitioning of public transit networks is a useful approach to speed up the query times or the preprocessing phase of public transit routing algorithms. The aim of partitioning of public transit networks is to find small partitions of the stations of a given network such that most of the traffic lies in their inside while only little traffic goes between them. To find such partitions of the public transit network of Germany, the public transit network is modeled as a graph, where nodes represent stations and edges are weighted with the frequency of vehicles going between the stations. With k-means-clustering, a merging-based algorithm, partitioning using natural cut heuristics (PUNCH) and METIS four graph partitioning algorithms are tested and evaluated on a dataset that contains the schedule of Germany of the year 2015. It turns out that the merging-based algorithm and METIS have fast runtimes and produce small partitions with a small cut size. K-means-clustering solves the problem better than expected. PUNCH was invented for unweighted graphs and needs some adaptations for weighted graphs.