The increasing number of traffic accidents worldwide necessitates innovative approaches to enhance road safety. This study integrates machine learning techniques into traffic management systems to predict accident severity, utilizing historical data from the UK Government’s open data portal (2018–2022). The dataset includes 552,648 rows and 24 columns. Various machine learning models were evaluated to determine their effectiveness in predicting accident severity. The Random Forest classifier demonstrated superior performance across all metrics, correctly identifying 80% of the labels across all classes. Data preprocessing involved merging vehicle and collision datasets, removing non-essential columns, and handling missing values to ensure data quality. Feature engineering categorized engine capacities and generalized accident timestamps to enhance predictive power. The study highlights the potential for real-time implementation of these predictive models in Intelligent Transportation Systems (ITS), providing actionable insights for driver guidance and improving road safety. Future research could integrate real-time data and advanced ML frameworks like AutoML to refine these models further, making traffic management strategies more dynamic and responsive to real-world conditions.
Enhancing Road Safety: Predictive Modeling of Traffic Accident Severity Using Machine Learning
Smart Innovation, Systems and Technologies
World Conference on Information Systems for Business Management ; 2024 ; Bangkok, Thailand September 12, 2024 - September 13, 2024
21.06.2025
11 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Classification of Traffic Accident Severity Using Machine Learning Models
Springer Verlag | 2025
|