Autoregressive Integrated Moving Average (ARIMA) modelling and Feedforwar Neural Network (FFNN) with Backpropagation algorithm untuk predict Open Price of world Gold on July 2008 - February 2014.

Abstract

The objective of this study is to describe Autoregressive Integrated Moving Average (ARIMA) modelling and Feedforward Neural Network (FFNN) by Backpropagation Algoritm from open price of gold data, seeing parameter estimation mode of ARIMA and FFNN, predicting open price of world gold data for the next one month, and seeing the accuracy of world gold open price data on July 2008-February 2014 on the result of ARIMA compared with Feedforward Neural Network (FFNN) algoritm. Based on identification process of ARIMA model used in this research is model of ARIMA (1,2,1). And for data estimation of gold open price using Feedforward Neural Network (FFNN) with backpropagation algoritam method an dthen results optimum FFNN model. Estimation result of world gold open price on March 2014 using ARIMA (1,2,1) model is adalah about US$1180,55 per troy ounce with the score of MSE about129,89%. And the estimation result of world gold open price on March 2014 using  Feedforward Neural Network with backpropagation algoritm is about US$1011,7 per troy ounce with MSE score of 28,281%. From both methods  used for visible estimation result of ARIMA model (1,2,1) results bigger estimation score. However, the MSE score of Feedforward Neural Network with backpropagation algoritmrelatif is fewer than MSE ARIMA score (1,2,1), that shows accuracy level of estimation result with backpropagation algoritm method is better than ARIMA (1,2,1).