Abstract: With the rapid advancement of smart grid construction, electricity information acquisition systems have accumulated massive volumes of user electricity consumption data. Accurately fitting user electricity consumption patterns from such data has become a critical task in power system data analytics. As an ensemble learning method, the random forest algorithm offers advantages such as strong feature selection capability, resistance to overfitting, and robustness to outliers, making it suitable for handling complex datasets in electricity information acquisition systems.This paper presents a systematic implementation of the random forest algorithm for electricity consumption fitting, including key stages such as data preprocessing, feature engineering, model construction, and performance evaluation. Case studies are conducted to validate the effectiveness and superior performance of the proposed method in electricity consumption fitting applications.