Comparison of Machine Learning Algorithms for Predicting Traffic Accident Severity

Abstract

Traffic accidents are among the most critical issues facing the world as they cause many deaths, injuries, and fatalities as well as economic losses every year. Accurate models to predict the traffic accident severity is a critical task for transportation systems. This investigation effort establishes models to select a set of influential factors and to build up a model for classifying the severity of injuries. These models are formulated by various machine learning techniques. Supervised machine learning algorithms, such as AdaBoost, Logistic Regression (LR), Naive Bayes (NB), and Random Forests (RF) are implemented on traffic accident data. SMOTE algorithm is used to handle data imbalance. The findings of this study indicate that the RF model can be a promising tool for predicting the injury severity of traffic accidents. RF algorithm has shown better performance with 75.5% accuracy than LR with 74.5%, NB with 73.1%, and AdaBoost with 74.5% accuracy.