While meeting with one of our partners and one of Latvia’s largest passenger transport companies JSC “CATA” the necessity for bus arrival time prediction system was identified. The system developed by RTU ITI is based on opensource technologies (PostgreSQL, PostGiS, scikit-learn, NodeJS, Express, VueJS). It is able to import public transport information in GTFS format and train machine learning models based on accumulated historical data. Once machine-learning models are present, the system matches individual buses to routes and trips and evaluates their arrival times with the help of the machine learning models.
Figures below contain an overviiew of the solution architecture and example of the user interface.