Credit evaluation of electric clients based on SVM

Combined with the basic situation of China's electric clients,a credit evaluation index system for electric clients was established.The quantitative indexes were pretreated with non-dimensional treatment and the linguistic variables describing the qualitative indexes were presented with trapezoid fuzzy numbers.Then,the trapezoid fuzzy numbers were converted into clear non-fuzzy value by using its μ-expected value.Finally,the electric clients credit was evaluated by means of SVM and realized by OSU_SVM3.00 SVM toolbox in the environment of Matlab6.5.Experimental results show that this is an effective method to evaluate the electric clients credit.