Road Traffic Safety Prediction Based on Improved SVM

Abstract

Traffic safety is considered as one of the important problems in the development of cities. Thus, a reliable estimation of traffic safety is crucial to be proposed. As predicting outcome indicators, two relative indicators, accident rate per 10000 cars and accident rate per 10000 capita, are introduced in this paper. Based on the theory of support vector machine (SVM), the improved SVM models are proposed. In order to improve the efficiency of learning, the least support vector machine (LSSVM) is introduced; to improve the model's robustness further, the weighted least support vector machine (WLSSVM) is introduced. As an example, using the traffic safety data which collected from 2006 to 2010, the models' With the swift development of road traffic industry and automobile industry, road traffic safety issues have aroused more and more concerns from the society. Road traffic accidents cause not only casualties but huge economic losses, and even bring about a series of social problems. In road traffic safety prediction, the data relevant to traffic safety will be analyzed to find out the frequency of road traffic accidents and reasonably estimate the possibility of future accidents. This will be helpful to the improvement of urban traffic safety management level, to accident prediction and prevention management and to the acceleration of the systematization, scientification and modernization of urban road safety management(Zhang