The Application of SMOTE Algorithm for Unbalanced Data

Abstract

The current power user data is unbalanced when it is used to analyze the behavior of the leakage user. In other words, the normal user data and the leakage user data have an inconsistent scale. When the automatic identification model of the leakage user is established, the analysis of the information of the leakage user's behavior feature is not clear, which leads to the reduction of the model's efficiency of classification. In this paper, we use Python and deal with the leakage user data based on SMOTE algorithm to increase the basic information of the users and extract more accurate leakage user behavior characteristics.