Abstract:In order to solve the difficulties in recycling, the difficulties encountered in recycling, the difficulty in recycling, and the incomplete recovery of electricity charges have seriously affected the final operating results of the power supply enterprises. This paper combines the data generated by various users in various production activities in the past year to establish models related to user behavior. Firstly, the clustering method is adopted, and according to the clusters finally obtained, there is a general understanding of the types of electricity charge recovery risks of all users in Chongqing. Then the logistic regression model is adopted, and the three different types of users of high-voltage, low-voltage residents and low-voltage non-residents are separately analyzed and modeled, and the score of the user''s electricity fee recovery risk is finally obtained. This paper proposes a probabilistic clustering logistic regression model for the electricity cost risk prevention and control business, which is used to realize the refinement of the arrearage risk. Analyze and locate customer groups, refine, summarize, summarize customer characteristics, and achieve customer segmentation. Provide accurate and differentiated services based on customer segmentation results. Practice has proved that this model can provide effective support for the practical application of risk assessment for grid enterprise customers.