SwissRailNet: Four representations of a spatially embedded railway network dataset.

Public transport networks are frequently studied in the field of network science, including applications in epidemic modeling, infrastructure optimization, and machine learning. This paper introduces four representations of a Swiss railway network. The dataset is derived from open data sources of Swiss public transport companies and is the result of several data processing steps as well as a manual validation. In all four representations, nodes represent train stations, embedded in geographic space, making the network an example of a spatial network influenced by physical constraints such as mountains and lakes. The edge definitions vary, however. Edges connect (1) any pair of nodes with a train connection for which no train change is required, (2) any pair of nodes with a non-stop train connection, and (3) any pair of nodes that correspond to physically adjacent stations. The fourth representation corresponds to a temporal network in which each edge is attributed with the start and duration of a train connection.