The Occupancy API : how does it work?

The Occupancy API works by analyzing the usual crowding of courses based on past counting cell data.

Usual crowding analysis

It relies on several data sources:

  • Système d'aide à l'exploitation (SAE) or another intelligent transport system (ITS) that can provide past course history with real and scheduled departure and arrival times, vehicle information, and course unique identification;
  • Counting cells data (or other APC systems);
  • GTFS or another transport referential format.

Using these data sources, CITiO's software systems can reconstruct past occupancy for all courses with several levels of aggregation:

  • line,
  • direction,
  • stop,
  • time (by 15-minutes periods).

Prediction

The Occupancy API uses 6 months of historical crowding data to compute the average occupancy of a scheduled course and expose it in real-time.

It can predict occupancy based on denoised counting cells data, ticketing data, or a combination of both (unified occupancy, documentation in French).

The type of occupancy can be configured by line. If the selected type is occupancy, values can be adjusted through a coefficient to reflect known fraud rates.

Then the occupancy and capacity are averaged by:

  • line, direction and route
  • stop
  • day of the week
  • 15mn time period
  • period of the year: workdays or school holidays

Assuming new data is sent to Citio on a daily basis from all data sources (except GTFS or equivalent), the predictions are updated nightly.

Prediction are adjusted on a weekly basis, using a seasonality coefficient accounting for seasonal ridership variations. For example, cities with a large student population will have a higher ridership in October than in June.

Occupancy API provides predicted occupancy data for any day in the future.

Go back to API main page.

Avez-vous trouvé ce que vous cherchiez ?

What is the Occupancy Prediction API?

The Occupancy Prediction API : how to use it?

Contact