抄録
This study adopts artificial neural networks and multiple regressions models with two weeks explanatory variables of daily exchange rates feature to analyze the accuracy forecasting technique. The training data set is used to compare the capability of prediction function to explain a correlation between predictor variable and response variable. The result indicated that forecasting exchange rate by artificial neural network is better in explaining a correlation between predictor variable and response variable compared to multiple regression. The statistical test indicated that forecasting exchange rates by artificial neural network is more accurate significantly compared to multiple regression.
