Abstract

Road accidents remain a leading cause of death, disability, and hospitalization in the country, highlighting the need for effective traffic accident risk prediction to save lives. Various models, from traditional statistical methods to modern machine learning techniques, have been developed for this purpose. This paper compares these different models to identify the most effective approach for predicting traffic accident risks. Since drivers have control on the road, the study focuses on providing them with risk predictions based on factors they can anticipate, such as vehicle type, age, gender, time of day, and weather conditions. Models like Optimal Classification Trees, Random Forest, and Logistic Regression offer results that are easy for drivers to understand. Additionally, analyzing geo-location data with K-means clustering can identify accident-prone areas. By using these algorithms to analyze known factors, drivers can receive traffic accident risk predictions, helping them make informed decisions to reduce accidents. The project's goal is to enhance traffic prediction and road accident analysis, ultimately improving road safety.