Data mining on uncertainty is concerned with. Support Vector Machine (SVM) is a powerful classification technique and has been successfully applied to many real-world problems. However, it is required that the values of input indices must be real numbers. A new fuzzy SVM technique is proposed which can deal with the training fuzzy data. A definition of fuzzy coefficient quadratic programming is given as a direct result of fuzzy SVM. Meanwhile, a useful method is presented to solve the programming. By data experiment, ordinary SVM can be viewed as a special case of fuzzy SVM.