Demand Analysis and Occupancy Prediction in Bilbobus
Summary of the entity:
Bilbao tops the ranking of the main Spanish cities for its commitment to sustainable mobility thanks to an increase in the pedestrian zones, enforcement of public transport and deployment of cycle paths, according to a study conducted by Greenpeace.
Summary of the challenge:
The challenge objective would be to analyse the daily and recurring demand of buses to infer O/D matrices based on mobility patterns, in order to achieve a daily prediction, as well as in real time, of the occupancy levels of the buses.
In the post-covid era, the occupancy level of urban buses is a relevant criterion for many potential users, and both public administrations and transport operators want to analyse the occupancy levels of their buses and, if possible, publish these occupancy levels to the users.
This circumstance is certainly difficult in those services where the user only registers his entry into the vehicle and does not do so when leaving it, therefore the use of ticketing information can be key to offering this service to citizens.
- O/D matrix inference algorithm:
– Output: Daily O/D Matrices at expedition level. This is used to infer get-offs in each stop and the consequent bus occupation
- Transport Occupancy Prediction based on Deep Learning:
– Output: Using all the mentioned inputs and the occupancy data, the output should be a prediction of future occupancy at three levels: Next days, next expeditions (current day), and next stops (current expedition).