| Summary: | 碩士 === 國立高雄應用科技大學 === 國際企業研究所 === 102 === With the prevalence of international trade liberalization, capital flows between countries are getting frequent, and thus highlight the important of the study on exchange rates. If the variation of exchange rate can be captured, we will provide domestic investors a reference for better investment strateges, and help promote domestic economic development.
In this research, the forecasting results obtained by conventional time-series models and by the Interactive Artificial Bee Colony (IABC), which is a young artificial intelligent method, are compared with each other with 8 years historical data. The sliding window strategy is used in the experiment for both the training and the testing phases. In our experiments, we use continuous previous thirty daily data as the training set, and use the training result to forecast the foreign exchange rate on the next day. In addition, we evaluate the forecasting accuracy with one criteria, namely, Mean Absolutely Percentage Error (MAPE).
The experimental results indicate that feeding macroeconomic factors to IABC as the input data is capable to produce higher accurate data in the foreign exchange rate than the conventional time-series models.
|
|---|
