Energy Consumption Forecasts by Gradient Boosting Regression Trees

Recent years have seen an increasing interest in developing robust, accurate and possibly fast forecasting methods for both energy production consumption. Traditional approaches based on linear architectures are not able to fully model the relationships between variables, particularly when dealing with many features. We propose a Gradient-Boosting–Machine-based framework forecast demand of mixed customers dispatching company, aggregated according their location within seven Italian electricity market zones. The main challenge is provide precise one-day-ahead predictions, despite most recent data being two months old. This requires exogenous regressors, e.g., as historical features part air temperature, be incorporated scheme tailored specific case. Numerical simulations conducted, resulting MAPE 5–15% zone. Gradient Boosting performs significantly better compared classical statistical models time series, such ARMA, unable capture holidays.