Ensemble learning for electricity consumption forecasting in office buildings

Abstract This paper presents three ensemble learning models for short term load forecasting. Machine has evolved quickly in recent years, leading to novel and advanced that are improving the forecasting results multiple fields. However, highly dynamic fields such as power energy systems, dealing with fast acquisition of large amounts data from sources taking advantage correlation between available variables is a challenging task, which current not prepared. Ensemble bringing promising this sense, as, by combining use learners, able find new ways be used optimized. In developed respective compared: gradient boosted regression trees, random forests an adaptation Adaboost. Results electricity consumption hour-ahead presented using case-study based on real office building. show adapted Adaboost model outperforms reference