A Prediction Method of Anti-electricity Stealing Based on Big Data of Electric Power

This paper introduces the diversification and harmfulness of power theft methods in recent years, then collects and analyzes the power theft data of a power company from the power information collection system and the marketing business application system, and establishes the anti-theft characteristic orientation scale by using big data analysis method, and then forms the anti-theft expert sample data. By using the optimized mathematical model. This paper analyzes the principle of machine learning classification algorithms such as decision tree, support vector machine, K nearest neighbor and BP artificial neural network, select the optimal algorithm, train and model the sample data, and write a program to predict the new data. MATLAB tool software is used in the experiment, and the practical model of the optimal classification machine is finally formed by constantly adjusting the classification algorithm and parameters. Through cross-experiment verification, the optimal classification prediction results are obtained, which can effectively improve the accuracy of electricity theft recognition.