Exchange Rates Forecasting and Comparative Analysis Based on Neural Networks

Exchange rates float with non-linear dependence and long-term memory characters, so we choose the neural network tool to forecast it. Based on the previous research work, we use two groups of data at different time-intervals to make forecasting experiments with homogenous neural network model and heterogeneous neural network model. According to the results of the experiments, we make a comparative analysis and find suitable application areas of these two models.