Parameter optimization of Brown's and Holt's double exponential smoothing using golden section method for predicting Indonesian Crude Oil Price (ICP)

Abstract

Oil fuel is a vital necessity for the people of Indonesia. Government policy that diverts subsidy and transfers the prices to economic mechanisms can make fuel prices fluctuate irregularly so that it can bring startled to the public if it is not anticipated with preventive measures. One indicator influencing fuel prices in Indonesia is the Indonesian Crude Oil Price (ICP) so that the forecasting mechanism to predict ICP becomes important. The exponential smoothing method is one of the time series prediction models that can be used to predict ICP. The problem arising in this method is to determine the optimal parameters to minimize the forecast error. This research focuses on the parameters optimization using golden section method. Data used in this research are ICP of January 2011 until May 2016. The analysis showed that the data is trend patterned so it is appropriate to use double exponential smoothing (DES) method from Brown and Holt. Data is divided into training and testing data in the ratio of 80:20. Residual between the prediction results using optimal parameters and testing data used to test the feasibility by performing normality test and randomness test. The results of parameter optimization are the optimum value of α in DES Brown is 0.47206 and the optimum MAPE of 13.061%, while in DES Holt the optimum α is 0.56341 and the optimum γ is 0.05463 with the optimum MAPE of 13.063%. Feasibility studies showed that both methods are feasible for prediction. DES Brown was selected as the best model for the prediction based on the value of MAPE and feasibility studies.