A spatio-temporal network of the New York Subway. This is a spatially embedded time-varying public transport network. The data file is a gzip compressed tar archive containing two CSV files. The network represents timetabled New York Subway journeys for two days. Specifically, 2015-11-30 00:00 to 2015-12-02 00:00 (EST). "newyork_stnet.csv" describes the time-evolution of the network's topology. This includes the time-varying transit speed between two locations. The file is a CSV file where each row describes a weighted directed edge in a particular timestep. Each line consists of the following fields: – node1: Source node of the directed edge. – node2: Destination node of the directed edge. – layer1: The index of the timestep that the edge appears in. Runs from t1, t2, t3, …. – layer2: Always the same as layer1. (Can be ignored.) – timestamp: An ISO-format datetime string describing the start time of the snapshot that the edge appears in, including timezone offset from UTC. – weight: The propagation speed from source node to destination node during this timestep. Measured in metres per second. A non-zero value. The absence of a row for two nodes in a particular timestep indicates that there is no edge between them during the corresponding time interval. In addition, geographic locations of nodes are provided in each dataset's coordinates CSV file ("newyork_coords.csv"). Columns consist of node, longitude, and latitude. Longitude and latitude are given on the WGS-84 ellipsoid. Node locations in these two datasets are not time-varying and therefore their coordinates are not associated with a particular timestamp. This dataset is derived from the New York City Metropolitan Transportation Authority (MTA) GTFS schedules available through the MTA static data feeds.