Forecasting Gold price with Hybrid Model

The econometric models such as ARIMA (Autoregressive Integrated Moving Average) are highly capable to predict the linear data of time series, while the neural network is highly capable to predict the nonlinear models. In the real world, determining the data to be linear and or nonlinear is difficult, thus using the combination of the two linear and nonlinear models to predict time series result in improved precision. On the other hand, the financial data has usually noisy and this will reduce the precision to predict the models. Hence this study compared the forecasting accuracy of hybrid models using raw and de-noised data for weekly gold price per ounce. Experimental result show that hybrid model with de-nosing data outperform the hybrid model with raw data.