In order to assist the electricity sale company to improve the quality of the market-oriented sale service of electricity, support vector machine-chaotic time series model is used to dynamically predict the electricity consumption of users who can participate in market oriented electricity sales in Shaanxi. Then, users who do not participate in the marketization are analyzed and evaluated according to the calculation results of the deviation of electricity consumption forecast and the decrease of cost. Agent suggestions are given for electricity sale companies to carry out sale business. Finally, according to the requirements of the assessment, the dynamic prediction deviation value of the users who have partici-pated in the marketization is used to further analyze whether the users are assessed or not. The preliminary application results show that the electricity sale company has achieved good results in carrying out agency business from these two aspects, and the application can be further deepened.