Abstract
Clustering of buses along a route is a major prevailing issue in the public transport system where multiple buses arrive at bus stops in quick successions rather than keeping scheduled time gaps in between. This phenomena has many negative impacts on the reliability and efficiency of public bus transportation system. Increased passenger waiting times and uneven distribution of passengers among buses, leads to overcrowding in some buses and will lead to passenger discomfort which ultimately influence the general public to avoid public transport altogether. This research aims to predict stop-level bus clustering events using a combination of bus data from Bus Open Data UK and weather data.Route 229, a busy corridor between Huddersfield and Leeds, was selected as the case study. The dataset includes over 112,000 stop-level records enriched with engineered features such as delays, scheduled gaps, and temporal indicators. Ensemble machine learning algorithms, Random Forest and XGBoost were trained using a stratified and balanced dataset, with performance evaluated through cross-validation. The models achieved high predictive accuracies. The results indicate not only the feasibility of using open-access bus and environmental data for clustering prediction, but also the effectiveness of the proposed modelling approach in capturing and forecasting stop-level bus clustering patterns.
| Date of Award | 19 Aug 2025 |
|---|---|
| Original language | English |
| Supervisor | Mauro Vallati (Main Supervisor) & Simon Parkinson (Co-Supervisor) |