Incorporating decision-making styles to predict driver-injury severity in road accidents in a large metropolitan area: a machine-learning-based approach

Traffic accidents around the world cause significant economic, human, and social losses annually. As a result, they have always involved their own macro policies and executive plans. Proper planning in this area requires a thorough understanding of traffic accidents. Identifying and analysing the causes of traffic accidents help make better predictions about them and the severity of their injuries. In addition to the well-cited reasons such as vehicle and road conditions, this study explored driver's decision-making style as one of the factors affecting the severity of traffic accidents. The purpose of this study was to predict traffic accidents and the severity of their injuries by considering the decision-making style of drivers. To this end, we developed and analysed different scenarios according to a variety of data sorting modes, data pre-processing methods, and various classifiers based on machine learning. The results showed that considering the decision-making style has a positive impact on the performance of the prediction model. It was also found that the best-case scenario occurs under the following conditions: 1) all the data alongside decision-making style are presented to the model; 2) outliers are excluded in a permissive mode; 3) the AdaBoost classifier is used for making predictions.