Abstract: Accurate prediction of energy consumption is helpful for further value mining and data fusion. In order to achieve this purpose, this paper proposes an energy consumption prediction method based on SARIMAX (seasonal autoregressive integrated moving average with exogenous) and XGBoost (eXtreme Gradient Boosting algorithm) hybrid model. First we import the training data required for the experiment and the auxiliary weather environment data, compared curve relationships and the correlation coefficient matrix, and used k-means to build weather clusters. Then we built holiday indicators, made further adjustments based on seasonal trends, and used grid search to select the optimal parameter combination of the SARIMAX model. Finally, we fused XGBoost algorithm to optimize the prediction model, made predictions and compared the implementation results. Through the analysis of the results, it can be seen that the hybrid SARIMAX model and the XGBoost model can accurately predict regional energy consumption on the basis of considering multiple exogenous variables.