Prediction of road crash attributes using machine learning

Abstract

Road crashes are a significant global problem, resulting in the loss of over 1.19 million lives and causing around 50 million severe injuries each year. However, despite the efforts made by the Government to improve the situation, no significant results have been observed in respect to decline in number of accidents and deaths in India. Road safety is a complex issue and requires a multifaceted approach including adequate research and development in the field. At present the mostly traditional approach is applied in terms of data collection, statistical analysis and decision making. Over time, the Government may start building advanced data analytics solutions for prediction of various road crash attributes. With this intent this paper aims to depict the application of combination of different algorithms including Support Vector Machine, Random Forest, Decision Tree, Artificial Neural Network, K Nearest Neighbor, Logistic Regression, and Gradient Boosting Algorithm for prediction of road crash attributes.