Project Summary
This project aims to create bus forecasts by developing a Machine Learning Model, calculating potential delays based on features such as weather and time of day, allowing us to better inform travellers in earnest.
Project Achievements
Activities • Specification and development of a predictive model for bus arrivals and departures, based on data from DfT BODS. • Qualitative user research into personalised messaging and alerts for bus services through the existing Zipabout platform. • Design and initial implementationof a real- world demonstrator allowing for quantitative research into messaging effectiveness.
Conclusions
• Successfully created simple predictions based on DfT BODS SIRI-VM and GTFS data for the majority of services. • Analysis for applicability of Random Forest Machine Learning (ML) model to bus predictions undertaken – previous approach for rail delays shown to be valid. • Country-wide deployment of technology with Oxfordshire Country Council (OCC) confirmed – supporting the Local Authority with net-zero targets and commitments.
Next Steps
• Further development of ML model with UoB through wider R&D partnership. • Launch of Zipabout Local product with OCC in Q4’22, enabling further research into effectiveness of predictions, and iterative development of solutions. • Integration of new prediction capability for bus services into Zipabout’s existing, national deployment with major rail operators (including National Rail Enquiries).
