Skip to content

Latest commit

 

History

History
71 lines (54 loc) · 2.57 KB

README.md

File metadata and controls

71 lines (54 loc) · 2.57 KB
Raw

hsl-congestion

Estimating Helsinki public transport congestion based on stop times.

How does it work?

There's a separate script, hsl-congestion-recorder, tracking trams through the HSL Realtime API. Every time a tram opens or closes it's doors, the event is registered to a database.

This service calculates a congestion rate from the door status events. First, it calculates how long the doors were open at the last stop. Then, how long they were open at the previous stop, and the previous, until it has the stop times for all the stops on the trip.

Next it sums up the stop times, but divides each stop time with the amount of stops that have come after it. Full weight is given only to the latest stop, half weight to the 2nd to latest, third weight to the next, etc.

Lastly it divides the resulting number with the average stop time for the corresponding part of the trip (ie. which stops have been passed), based on all previous trips on the route. This results in a nice ratio, where:

  • 0.0 means the tram is empty.
  • 0.5 means the tram is half as full as usual.
  • 1.0 means the tram has average congestion.
  • 2.0 means the tram has double the average congestion and is probably quite full.
  • etc.

Demo

  1. Find the GTFS ID of any active tram trip.

    Easiest is to go to https://api.digitransit.fi/graphiql/hsl.

    For example, for trams arriving at Kampintori from the direction of the central railway station, you can do:

    {
    stop(id:"HSL:1040409") {
        stoptimesWithoutPatterns {
            trip {
                gtfsId
            }
        }
    }
}

    Pick one of the returned trip GTFS IDs.

  2. Request the congestion rate endpoint with the trip GTFS ID.

    The endpoint is https://hsl-tram-congestion.appspot.com/trips/:tripId/congestionRate.

    For example,

    $ curl https://hsl-tram-congestion.appspot.com/trips/HSL:1007_20190820_Ti_2_2101/congestionRate
0.5287651445469622

    The returned number, ~0.52, ie. half of normal congestion, is the live congestion rate for the specified trip!

Contributing

  1. Clone this repository to your local machine.
  2. Run npm install
  3. Configure the database (and optionally SSL) settings into config/[NODE_ENV].json. The default NODE_ENV is development, so you should create a config/development.json.
  4. Start the development server with PORT=3000 npm run dev. If you want to run it in a different port number you can change the PORT argument freely.