本研究以灰色預測模式建立臺灣地區國際航空總客運量及來臺旅客與國人出國運 量預測模式。研究建立之灰色模式包括GM(1,1)時間數列預測模式及多變量GM(1,N)系統預測 模式。灰色預測模式不若傳統多元迴歸模式須要求大樣本量、較好的分布規律等條件。將原 始數據經灰色生成可轉化為生成數，生成數為灰色模式主要觀念及其微分方程組構建之基礎 ，且生成數具永續效果之意義。研究中所建立之灰色模式包括總運量預測模式及來臺旅客與 國人出國預測模式，由與多元線性及乘冪迴歸模式、ARIMA模式預測結果之比較分析，證實 以灰色模式建構之國際航空客運量預測模式確實比傳統迴歸分析模式與ARIMA模式較具解釋 與預測能力。總運量系統預測模式之預測結果與1991年至1995年實際運量值比較平均誤差為 1.78%，總運量時間數列預測模式平均誤差為4.42%，來臺旅客與國人出國預測模式1991年至 1994年平均誤差為3.04%。

　　　　　This paper applies grey theory to develop international air travel forecasting models for Taiwan Area. Several grey forecasting models (GM) including one time-series forecasting model of GM(1,1) and a variety of polyfactor forecasting models of GM(1,M) are presented. GM differ from conventional multiple regression models which demand large amount of data with good statistical distribution. Accumulation generating operation (AGO) functions are the essential concepts and the basis of grey differential equations. AGO imply permanent effects of affecting variables. Compared with the results of ARIMA models and regression models, we found that the models developed did produce less prediction errors while provide better economic insights. The results show that compared with recent five years (1991∼1995) actual data, the average prediction error is 1.78% for total passenger systematic forecasts, 4.21% for total passenger time-series forecasts, and the four years (1991∼1994) average prediction error is 3.04% for Taiwan resident departure and foreign visitor arrival forecasts.