Prediction Severity of Road Fatal Accidents Using Aggregative and Basis Classification Models

Due to the severity of accidents in suburban roads usually more than of the severe of accidents on city streets, so The prediction of road accidents in suburban routes is important .In this paper, we use the independent variables such as accident cause, collision type, fault vehicle type, accident day and accident season is investigated to predict the severity of road fatal accidents. For this purpose, a variety of basis classification models is based on such as classification K- Nearest Neighbor, Naïve Bayes, Decision Tree and Boosting and Bagging aggregative classification Using of 2008 fatal accidents data from national freeways in Iran have been fitted. The results from fitting models on the one hand have indicated more accuracy of Boosting and Bagging aggregative classification models in size of 0.04 and the other Bagging aggregative classification models with decision-tree learning algorithm and Boosting Aggregative classification with Naïve Bayes learning algorithm on prediction of road fatal accidents severity has had high performance.