Energy Consumption Forecasting using Ensemble Machine Learning Models–A Case Study Approach

Precise load forecasting of energy demand resulting in minimal / negligible variations in real-time load curve is crucial for effective operation and planning of utility companies. The prediction further allows efficient energy storage and it could provide useful information to make decisions on energy generation and purchase. Machine learning algorithms are one of the best accurate algorithms used in prediction. In this research work we implemented the ensemble methods, which are Ada Boost Regression (AdB), Gradient Boosted Regression (GBR), and Extreme Gradient Boosting Regression (XGB) were used to predict the energy consumption. The outputs of these models were evaluated to see which one provides the best energy prediction performance. The outputs indicate that simple ensemble learning techniques such as averaging expected results could help enhance prediction models. The results of these algorithms are compared with Root mean square (RMSE) and mean absolute error (MAE).