Application of Machine Learning and Deep Learning Techniques in Road Safety.

Abstract

Road traffic accidents are a growing global concern, causing around 1.3 million deaths and over 50 million injuries annually. This crisis highlights the urgent need for intelligent systems that can enhance road safety and reduce both the severity and frequency of accidents. This project investigates the use of ML and DL to analyze traffic accident data and support decision-making. It focuses on building predictive models that learn from many factors and past patterns to help prevent future accidents. Two key tasks are studied in this project: predicting accident severity using datasets such as US Accidents (2016–2023) and Road Traffic Accidents, and predicting accident frequency using Thailand Fatal Road Accidents and Great Britain Accident datasets. Each task used data suited to its objective. The project follows a specialized modeling process for machine learning and deep learning, known as CRISP-ML. A comprehensive data preparation steps were applied to clean and structure the data for effective training. By leveraging ML and DL techniques, several models are built and evaluated to demonstrate the effectiveness of these new approaches in enhancing road safety