Application of XGboost in Electricity Consumption Prediction
Danhuang Dong, Fan Wen, Yihong Zhang, Wenxin Qiu
Abstract
In this paper, a comprehensive power consumption forecasting method based on Xgboost algorithm is proposed to solve the problems of numerous influencing factors and different user behavior patterns in regional user power consumption forecasting. This method collects historical electricity consumption data and meteorological data to supplement the missing values in the data. The obtained user data is aggregated by K-means to obtain users of different categories. The maximum information coefficient (MIC) was used to calculate the correlation between each influencing factor and the power consumption of users. Xgboost algorithm is used to construct electricity prediction models for different user categories and then obtain the overall electricity consumption in the region. The user data of an industrial park in a province is used for experiment and compared with the results of random forest algorithm. Experiments show the effectiveness and superiority of XGBoost algorithm in short-term load prediction.