OptimasiPemulusanEksponensialdenganAlgoritmaLevenberg-Marquardt

Hery Andi Sitompul (Universitas Darma Agung)

Jurnal Darma Agung
Vol 26 No 1 (2018): AGUSTUS 2018

Determination of the value of the exponential smoothing’s parameters is a very sygnificant problems, because the value of these parameters is very sensitive to changes of forecasts in demand. Commonly, the way how to find the value of these parameters is try and error method. Try and error done by updating every possilble combination of the parameter’s values to exponential smoothing model thus, the results of the forecast is to be compared by each others. The best parameters is choosen by acuration measurement (minimum error),because of this method is very hard and wasting time,in this paper will propose another method how to get the best value of the smooting’s parameter. Non linear programming algorithm (in this case the levenberg – marquardt algorithm) will be aplicated to exponential smoothing model. In this paper will show that this algorithm is very fit to estimated optimum value of the smoothing’s parameters