
This study introduces a machine learning-based approach that integrates telematics data, OpenStreetMap (OSM) data, and official crash records to systematically categorize urban road segments in Athens as “Safe” or “Unsafe”. Driving behavior was quantified through the calculation of ratios for each key metric—speeding, harsh braking, and harsh acceleration—normalized against the number of trips per segment to mitigate the bias introduced by high-traffic roads. This normalization allowed the model to focus on relative driving behavior intensity rather than absolute counts. This research underscores the value of integrating data science and urban planning to address the growing challenge of urban traffic safety. By bridging crash records with telematics-derived behavioral data and applying machine learning, the study delivers a practical tool for evidence-based decision-making. The results not only contribute to safer road design and traffic management in Athens but also provide a replicable methodology for other urban centers seeking to adopt smart and proactive road safety strategies.
ID | pc593 |
Presentation | |
Full Text | |
Tags |