Abstract:For a long time,the problem of electricity stealing has been plaguing power enterprises.It not only detriments the legitimate rights and interests of power enterprises,disturbs the normal order of the power supply,but also causes the electrical safety threat.The data of electrical power with machine learning algorithms is analyzed,which can predict the existence of users stealing power behavior.Based on feature extraction of electricity consumption in power data,and by using the random forests algorithm,a method of predicting the existence of users stealing power behavior is proposed.By comparing multiple sets of experimental data,the parameters of the algorithm are adjusted to improve the accuracy of forecasting.