Abstract:Customers are not only the greatest wealth of enterprises, but also the biggest source of risk. Default of electricity charges, breach of contract and other behaviors occur from time to time. In the course of transaction, it is very important for enterprises to collect and investigate customer reputation information, to make risk assessment, and to formulate targeted prevention and service strategies. In order to reduce the risk of electricity tariff recovery and improve the business efficiency and management means of enterprises, customers need to be evaluated and graded. Electric power enterprises can use existing resources and data to establish customer credit information files, provide customers with differentiated services and avoid risks. Based on the huge amount of historical electricity data of customers in the marketing business application system, then draw lessons from the work experience of electric personnel such as customer managers and fee collectors, comb out the business rules of risk users, and build a multi-dimensional electricity risk analysis index system. The potential relationship between power consumption behavior and electricity risk was studied through the data analysis method of machine learning, and the precise location of risk users was realized.