An XGBoost Algorithm Based Missing Value Filling Method for Power Data

With the development of data science and computer technology, data processing has also become an essential part of our life, work, and research, however, the increase of data storage will inevitably lose some data. In this study, for the problem of missing electric power data, a data missing value filling method based on the XGBoost (eXtreme Gradient Boosting) algorithm is proposed by utilizing Python language. The method predicts and fills the missing data by analyzing and processing the electric power data and using the XGBoost algorithm. Firstly, the missing eigenvalues are divided according to the characteristic importance of the eigenvalues in the target sample set mainly based on electric power data, and then for the missing samples under each eigenvalue, the XGBoost model is constructed for the prediction and filling of the missing values, and finally, the complete sample set is obtained. The results show that the data missing value filling method based on the XGBoost algorithm can predict the missing values more accurately and improve the accuracy and efficiency of data filling.