This repo contains code related to collecting and analyzing vehicle location data provided by Capital Metro. There are some specific metrics for every route that I'm interested in such as:
- Deviation of actual arrival times to scheduled times
- Time between buses at a stop (headway)
- Estimating the expected arrival window of a bus at a stop determined by a confidence level (i.e. 95% confidence interval)
The captured vehicle location data for each day is made available the next day on the data branch under data/vehicle_locations (shortcut). The data is available in CSV (Excel-friendly) and JSON (dev-friendly) formats.
Please take the data, do whatever you want, and share it.
At the moment, only vehicle positions are recorded.
Note: Most of the optional fields of the VehiclePosition message in the GTFS-RT spec are omitted in CapMetro's implementation. Here's what their data looks like:
| field | description | GTFS-RT reference |
|---|---|---|
| vehicle_id | ID of the transit vehicle | VehicleDescriptor |
| speed | speed of the vehicle when position was recorded | Position |
| lon | longitude of vehicle when position was recorded | Position |
| lat | latitude of vehicle when position was recorded | Position |
| route_id | ID of the route the vehicle is assigned to | TripDescriptor |
| timestamp | Moment at which the vehicle's position was measured | VehiclePosition |
| trip_id | Refers to a trip from the GTFS feed | TripDescriptor |
At the moment, there's the Go daemon which runs continuously on a set interval and sleeps longer when every route is inactive. CapMetro publishes a stream of real-time vehicle locations which the daemon captures and passes off to RethinkDB. The CapMetro specific code is located in agency/capmetro. The agency package contains code for parsing and fetching data from transit agencies, though only CapMetro is supported for now. The task package contains the code for running functions in their own goroutine on a fixed and dynamic interval.
bootstrap.py contains code for bootstrapping a new RethinkDB database with routes and stops from CapMetro's GTFS data.
reporting.py contains code for generating some GeoJSON data (example) from recorded vehicle locations, as well as some useful queries e.g. getting the historical travel time between two stops.