- Big Data
- Machine Learning
- Predictive Analytics
Delays, slowdowns, cancellations. Bother of commuters and more or less regular train passengers, and a daily challenge for t those who manage transport systems.
The Mahsfrog Data Lab has developed a Machine Learning (ML) model for the prediction of train timetables at the service of companies operating in the cargo and passenger transport industry on the Italian railway infrastructure.
The implemented solution trains ML models on historical travel data to make them capable of predicting, for a new travel, the time of arrival at the destination based on the train number, the departure and destination stations, and the scheduled start time of the travel.
In the experimentation phase, based on the continuous improvement logic, our Lab developed a model for predicting the time of arrival that reduces the Mean Absolute Error (MAE) by approximately 30% compared to traditional models.