由於全球經貿的自由化，各國之間經貿往來日益頻繁，相互的依存度也日益提高，使得各國廠商在進出口的業務上非常重視匯率的變動。以台灣而言，國際貿易向來是經濟發展的主要來源，而廠商近年來更積極從事國際行銷及/或運用國際金融工具從事海外投資計畫，對於匯率變動的掌握益形重要。然而自 1997 年 7 月東南亞爆發金融風暴以來， 20 個月內之匯率呈現劇幅的變動，使得進出口貿易商蒙受鉅額的匯兌損失，如何掌握匯率的趨勢而做出正確的決策絕對是進出口廠商面臨的嚴峻課題。本研究針對東南亞金融風暴期間新台幣兌美元匯率之歷史資料，建構短期的預測模式。以經濟部國際經濟統計資料庫 AREMOS/UNIX 自 1997 年 6 月到 1999 年 2 月間，共 87 筆週資料為實証之時間序列資料。首先以迴歸分析針對可能影響匯率之變數進行變數篩選，協助類神經網路預測模式之建構，而在類神經網路預測模式之建構方面，首先利用敏感度分析決定模式之參數，而在有關建構模式之穩健性評估方面，則利用訓練樣本佔總樣本之不同資料比例加以分析。實証結果顯示結合迴歸分析之類神經網路模式相較於迴歸分析、ARIMA 及類神經網路模式顯著有較好之預測能力。

The development of Taiwan's economy, as being an island in the Asian Pacific, heavily depends on international trade. And hence the profit margin of import/export traders is seriously impacted by foreign currency exchange rates. Fixed or stable exchange rate will definitely reduce the business risk of international traders. Historical Data demonstrates that NTD rapidly appreciated from 40:1 to 25:1 against USD since the Central Bank adopted the floating exchange rate policy in 1987. The exchange rate of USD tended to stay stable after the market became more and more mature. However, as the result of the financial crises originated from Southeast Asia, the exchange rate varied violently during that period and import/export traders suffered heavy losses. The purpose of this research is to present a US exchange rate forecasting model in integrating neural networks and regression analysis. Regression analysis is first used to extract important variables that may influence the US exchange rate. The obtained variables are then used as the input variables in building the neural network model. To demonstrate the effectiveness of our proposed method, the weekly USD exchange rate series from July 1997 to February 1999 was evaluated using the designed neural network model. Analytic results demonstrate that the proposed method outperforms the traditional regression, ARIMA and neural network models.