In addition, we find that the multiple regression method outperforms the time series method, based on the average forecasting accuracy. This finding shows that statistically sophisticated and complex methods do not necessarily provide more accurate forecasts
than simpler ones. With a MAPE of slightly more than 3% in both regions, it is shown that the multiple regression is suitable for infeed forecasting with a horizon of one year on an hourly basis. On the other hand, although the regression method provides lower MAPE compared to the time series method, it results in a larger maximum average percentage error. Hence, DSOs should choose among these methods based on the cost of available hedges in the financial market. If it is very costly to hedge for the extreme cases, then the firm may be better off with the time-series model. Otherwise, a regression models should be given preference.