近年來由於金融市場的自由化及政府減少對外匯市場的干預，新台幣匯率跟隨國際市場大幅波動、劇烈升貶的情形屢見不鮮。如何正確的預測外匯走向是一個非常重要的課題，尤其對於以國際貿易為主要經濟活動的台灣而言，其重要性更是不言可喻。如果政府、企業決策甚至個人的投資活動，能夠精確掌握匯率之變動而做出適當回應，便能減低因匯率變動所產生之風險。
近年來，有許多學者專家提出以模糊時間序列(Fuzzy Time Series)作時間序列的預測。現實生活中，一個事件可能會被多項因素所影響；因此，在處理預測問題時，考慮多項因素的預測模式會比只考慮單一因素的預測模式更為準確。本論文提出以加權模糊時間序列(Weighted Fuzzy Time Series)來處理匯率預測的問題。我們以匯率為第一因子，以主成分分析法(Principal Components Analysis)將其它影響匯率變數的線性組合當作第二因子，提出一個可調整第一與第二因子在決定匯率上的權重的加權模糊時間序列，經實驗發現，本方法可準確預測匯率及匯率變動的方向性。

Due to the liberalization of the financial market and the diminishment of the government’s intervention on the foreign exchange market, we have witnessed severe fluctuations of the exchange rates of the NT Dollars against different foreign currencies. Since the exchange rates of the NT Dollars against other foreign currencies have significant effects on the international trade of Taiwan, how to forecast the exchange rate variations becomes an important issue for Taiwan. If the government, an enterprise or an individual can accurately predict the exchange rate variations, then the capital loss due the exchange rate variations can be reduced.

Recently, many researchers have proposed to use the fuzzy time series to model and predict many real life time series applications, such as predicting university enrollments or daily temperatures. In this thesis, we propose a weighted fuzzy time series (abbreviated as WFTS) to predict the exchange rate of the NT Dollars against the US Dollars. We consider two factors in the proposed method. The first factor is the historical exchange rates of the NT Dollars against the US Dollars. The second factor is derived, through the Principal Components Analysis (PCA), of several variables affecting the exchange rates including the exchange rates of the trading competition countries and the Taiwan Stock Exchange Capitalization Weighted Stock Index (TAIEX). In the proposed method, we adjust the relative weight between the first factor and the second factor to find the better predicting rules to predict the future exchange rates. The experiment shows that the proposed weighted fuzzy time series model has a better forecasting accuracy rate compared to the random walk model and the FLAR model. Furthermore, the proposed method shows better directional symmetry than the random walk model and the FLAR model for predicting long term exchange rates.

1.Benjamin M. Tabak,“The Dyanmic Relationship between Stock Prices and Exchange Rates: evidence for Brazil”, Published in International Journal of Theoretical and Applied Finance, Vol. 9, no. 8, pp. 1377-1396 (2006)
2.C.C Nieh, and C.F.Lee,“Dynamic relationship between stock prices and exchange rates for G-7 countries”,Quarterly Review of Economics and Finance,41(4),477-490
3.C.C Tsai,and S.J.Wu,“Forecasting exchange rates with fuzzy logic and approximate reasoning”, Proceedings of the 19th International Meeting of the North American Fuzzy Information Processing Society, PeachFuzzy Altanta 2000.
4.Chakrabarti A“Import competition,employment and wage in US manufacturing：new evidence from multivariate panel cointegration analysis”Applied Economics, Volume 35, Number 13, 10 September 2003 , pp. 1445-1449(5)
5.L.X. Wang, and J.M.Mendel,“Fuzzy Sets”,Information and Control,Vol.8,pp. 338-353,1965
6.L.W. Lee, S.M. Chen, and Y.H. Leu,“A New Method for Handling Forecasting Problems Based on Two-Factors High-Order Fuzzy Time series”,Proceedings of the 2004 Ninth Conference on Artificial Intelligence and Applications, Taipei, Taiwan, Republic of China, November 2004.
7.K. Huarng,“Effective Lengths of Intervals to Improve Forecasting in Fuzzy Time Series,”Fuzzy Sets and Systems,Vol.81,No.3,pp.311-319,1996
8.Q.Song and B.S. Chissom,“Fuzzy Time Series and Its Models,”Fuzzy Sets and Systems, Vol.54,No.3,pp.269-277,1993
9.Q.Song and B.S. Chissom,“Forecasting Enrollments with Fuzzy Time Series- PartΙ,”Fuzzy Sets and Systems,Vol.54,No.1,pp.1-9,1993
10.Q.Song and B.S. Chissom,“Forecasting Enrollments with Fuzzy Time Series- PartΠ,”Fuzzy Sets and Systems,Vol.62,No.1,pp.1-8,1994
11.R.A. Meese,K. Rogoff(1983),“Empirical Exchange Rate Models of the Seventies:do They fit outSample？”Journal of International Economics,Vol.14,pp3-24.
12.S.M. Chen,“Forecasting Enrollments Based on Fuzzy Time Series,”Fuzzy Sets and Systems,Vol.81,No.3 pp.311-319,1996
13.S.M. Chen,“Forecasting Enrollments Based on High-Order Fuzzy Time Series,”Cybernetics and Systems:An International Journal, Vol.33,No.1,pp.1-16,2002
14.丁義恩(1998)，應用類神經網路在匯率走勢上的預測，碩士論文，義守大學。
15.邱至中(2001)，長短期匯率預測模式績效之比較，碩士論文，成功大學。
16.施向陽(2000)，匯率變動預測模式之研究，碩士論文，大葉大學。
17.陸君芬(1998)，有關匯率隨機漫步性質之再議，碩士論文，政治大學。
18.許嘉元(1993)，Fuzzy analysis and forecasting in time series，碩士論文，政治大學。
19.賴名諺(1991)，影響新台幣匯率因素-短、中長期匯率迴歸模型分析，碩士論文，中山大學。