The NeurIPS CityLearn Challenge (Forecasting Track) focused on predicting energy consumption and solar generation for buildings. Our solution combined statistical methods with machine learning approaches to achieve the top position on the leaderboard.