| dc.description.abstract | In the field of economics, scholars studied how to forecast exchange rates by economic models. In the field of computer science, scholars applied machine learning approach to forecast exchange rates. Although cross-disciplinary scholars often compare their empirical model with computer science models, they hardly compare the performance of economic forecasting models with the performance of machine learning approach. In this thesis, we applied Markov Switching Model, Vector Error Correction Model and Machine Learning approach to forecast the exchange rate of new Taiwan dollar. Besides, we compared the outcome of economic model with the outcome of machine learning models. The results show that, in the short run forecast horizon, there are insignificant difference between the economic models and the machine learning models. In the long run forecast horizon, there are significant differences between economic models and the machine learning models. | en_US |