Abstract: Traffic flow prediction is a very important area in intelligent transportation systems.Traditional prediction methods have a very wide range of applications in the traffic prediction.But traditional prediction methods does not work very well in short-term traffic flow prediction because of the complexity ofthe influencing factors.With the development of machine learning and data mining,traffic flow prediction with a combination of machine learning and data mininghas become more and more important as a research area.In this paper,SVM(Support Vector Machine) is used to build a short-term traffic flow prediction model,and Genetic Algorithm (GA) is used to optimize the SVM penalty factor C andkernel parameter σ as well.The results of different kernel functions of SVM arecompared,including polynomial kernel and RBF kernel.RBF SVM plays better thanpolynomial SVM with less training time and higher accuracy and SVM is very suitable for short-term traffic flow prediction.

Key words: SVM, traffic flow, short-term prediction, genetic algorithm