單變量ARIMA及類神經網路模式預測香港國際航空站旅客流量

航空運量對於航空產業的決策影響甚大，航空業通常以旅客流量多寡，規劃航線的機隊數量，藉此獲取企業最大利益。本研究以單變量ARIMA及類神經網路模式，預測航空旅客流量，並選用香港國際航空站為例，收集民國87年至99年之間，共計148筆月資料，進行模型測試。本文研究發現，ARIMA及類神經網路模型對香港國際航空站之預測MAPE值皆低於10％，顯示兩模式皆適合於預測應用，而以ARIMA季資料模型較佳。就預測結果而言，香港國際航空站的旅客流量未來仍是呈現向上的趨勢。本文研究結果應有提供航空業者於市場規畫之參考價值。

關鍵字

香港國際航空站；旅客流量；單變量ARIMA模式；類神經網路

並列摘要

The strategy management for the airliner is affected by the air traffic flow. Airliners have to plan their fleet routing based on the traffic flow to achieve maximum profits. In the paper, the ARIMA and Artificial Neural Network models (ANN) are used for air passenger flow forecasting with a real world case of Hong Kong airport as an example. The data for model test is obtained from year 1998 to 2010 with total 148 monthly data sets. The empirical result indicates that both ARIMA and ANN are suitable for Hong Kong airport forecasting with the value of low MAPE. For the forecasting result, the passenger flow in Hong Kong airport would tend upwards in the future. The paper can serve airliners as a helpful reference of their marketing plan.

並列關鍵字

ARIMA ； Artificial Neural Network ； Passenger flow ； Hong Kong Airport ； MAPE

參考文獻

香港特別行政區政府民航處網站數據資料。http://www.cad.gov.hk/english/home.html.

Google Scholar

Box, George E. P.,Jenkins, Gwilym M.,Reinsel, Gregory C.(2009).Time Series Analysis Forecasting and Control.John Wiley & Sons, Inc.

Google Scholar

Hagan, Martin T.,Demuth, Howard B.,Beale, Mark(1996).Neural Network Design.Thomson Learning, INC.

Google Scholar

Law, Rob(2000).Back-propagation learning in improving the accuracy of neural network-based tourism demand forecasting.Tourism Management.21(4),331-340.