Bus Classification method using GTFS data for Transport Mode Recognition: A call for Open Data

In the context of sustainable mobility, the ability to classify user behavior can support targeted interventions aimed at reducing greenhouse gas emissions. The proposed methodology builds on an existing framework capable of distinguishing between broad vehicle classes using GPS data, Human Activity Recognition, and GIS features from OpenStreetMap. While differentiating trips made by rail, bike, or road vehicles is generally feasible using accelerometer and location data, distinguishing between similar road-based modes, such as car, bus, and tram, remains a challenging task due to overlapping motion patterns and shared infrastructure. This paper presents a methodology based on GTFS information and user GPS data collected from a smartphone app running in the background. The results are promising with an F1 score of 68.7%, but scaling of the current implementation is slowed by the absence of open GTFS availability.