Robust Bootstrap and GARCH process application for heating demand in Germany

Energy, in all its forms, is a physical quantity without which man could not carry out all his daily activities. However, nowadays, we hear more and more about problems due to lack or waste of energy due to the misuse by man and also due to an exaggerated consumption of it. Making energy forecasts is therefore something that would allow man to balance what is required and consumed by energy without encountering problems of overproduction and shortage. To do this, increasingly sophisticated algorithms are created to allow companies that supply energy to stipulate plans for forecasting consumption, demand, and production of energy itself, allowing them to better manage economic and environmental resources. In the following, two main methods are reported: the classical residual bootstrap and the influence function bootstrap. In particular, the latter is a method capable of managing situations in which there are deviations from the assumed model and, with great satisfaction, the results obtained with this method have been better compared to the other confirming that, even in situations distant from the made assumptions, it is still possible to obtain more accurate future predictions.