Machine Learning to Predict the Freeway Traffic Accidents-Based Driving Simulation

The aimed of this research is to evaluate and compare different approaches to modeling crash severity as well as investigating the effect of risk factors on the fatality outcomes of traffics crashes using machine learning-based driving simulation. We developed prediction models to identify risk factors of traffics crashes can be targeted to reduce accident. The Random Forest model demonstrated the best performance from among the six different techniques with accurate 82.6%.