Extreme value theory-gumbel distribution with MEBoot algorithm on case of extreme wind speed

Forecasting methods are usually developed with two approaches that involve predictor variables and do not involve predictor variables. But if there are extreme values and limitations of data, then the accuracy of the forecasting model built by both approaches will decrease. Therefore, to identify the extreme value used Extreme Value Theory (EVT) method with Gumbel distribution approach. The advantage of this method is that it does not require predictor variables and can accommodate extreme values. In addition to overcome the problem of data limitations with the method of bootstrapping, so as to improve the accuracy of the model and the resulting estimator is robust. The study used EVT method with Gumbel distribution approach to identify extreme wind speed without involving predictor variable. The results of the analysis show that there is no change in extreme wind speed in the Semarang Statute but in December to February has the greatest extreme value potential compared to other months.