A Comparative Analysis of the Performance between Time Series Model and Back-Propagation Neural Network for Foreign Exchange

碩士 === 國立高雄科技大學 === 國際企業系 === 107 === Taiwan is an island-type country with a narrow population and few natural resources. The country is guided by international trading. Economic development must rely heavily on international trade, and capital flows between countries are quite frequent. In recent...

Full description

Bibliographic Details
Main Authors: CHEN,YI-YE, 陳奕燁
Other Authors: CHANG,JUI-FANG
Format: Others
Language:zh-TW
Published: 2019
Online Access:http://ndltd.ncl.edu.tw/handle/crs58a
Description
Summary:碩士 === 國立高雄科技大學 === 國際企業系 === 107 === Taiwan is an island-type country with a narrow population and few natural resources. The country is guided by international trading. Economic development must rely heavily on international trade, and capital flows between countries are quite frequent. In recent years, the amount of trade between Taiwan and china is the largest among countries in the world. Therefore, mastering the changes in the RMB exchange rate can provide reference for domestic investors and help to minimize foreign exchange risk. In this study, six kinds of economic variables affecting the exchange rate of RMB against the Taiwan dollar were used as research variables. The monthly data from January 2010 to December 2018 were used. Time series GARCH model and BPN were utilized to evaluate the performance of the forign exchange forcasting. The result shows that all variables have good performance in the prediction of the time series model. The prediction effect is the most accurate in the seven-year period, followed by the five-year period, and the three-year period is poor. Comparison between GARCH model and BPN, the BPN has better accuracy than GARCH model.