The Prediction of Rupiah Currency Toward US Dollar Using Hybrid Method

TESIS

Abstract

By : I Nyoman Sumerta Yasa

Email : nyoman_sumerta@yahoo.com

Faculties : Fakultas Teknik

Department : S2 Teknik Elektro

Forecasting is an integral part of management decision-making activities. Forecasts are made generally based on the data of the past that were analyzed by using certain methods. Therefore in this research will predict time series data using Radial Basis Fuction, ARIMA and Double Exponential Smoothing using Matlab version 8.1. The data used is the daily selling rate of Rupiah against the US Dollar that started from January 2012 to March 2014. Of the three results forecast, will be used voting method to obtain the accuracy of strengthening or weakening the condition of the rupiah against the US dollar and also the prediction results are combined by hybrid methods. From the results of RBF forecasting, ARIMA, Double Exponential Smoothing obtained MAPE respectively 0.66%, 3.32% and 0.94%, while the accuracy of the conditions to strengthen weakened by 52.54%, 45.76% and 52.54%. Of the voting result of strengthening and weakening conditions obtained an accuracy rate of 54.24% and after combined with methods hybridization obtained MAPE of 0.64% with an accuracy of 50.85%. It can be seen that for accuracy strengthened and weakened conditions are best obtained by using the method of voting, while for the best MAPE obtained by hybrid methods. Hopefully this research can help in analyzing the fluctuation of the value of a particular currency movements at the time of transaction - purchase of foreign currency.

Keyword : RBF, ARIMA, Double Exponential Smoothing, MAPE, Hybrid

Download : File 1 File 2 File 3