運用類神經網路建立國際旅客需求預測模型－以日本來台灣之國際旅客需求為例

Modeling and International Tourism Demand Forecasting for Taiwan：Using Artificial Neural Networks

李依璇(Yi-Hsuan Lee)

指導教授 ： 孫志鴻

國立臺灣大學/理學院/地理環境資源學研究所/碩士(2009年)

https://doi.org/10.6342/NTU.2009.10334

摘要

對於觀光產業來說，了解觀光需求是觀光分析的起點，藉由了解觀光需求可以了解觀光市場的喜好，以便觀光相關產業做決策的基礎，進而降低決策風險，因此觀光需求的預測在於旅遊管理中是非常重要的一環。並且研究指出影響觀光需求的因素有多種；然而，在先前的研究中，多以單一變數以及線性方程來預測觀光需求。因此本研究利用不同因素如(人口、匯率、GDP等)作為國際觀光客的需求的輸入變數，並且利用不同的類神經網路模型如自組性類神經網路(SOM)、輻狀基底函數類神經網路(RBF)、及倒傳遞類神經網路(BP)來建立國際觀光客的需求，並以日本對台灣的觀光客需求做為案例進行分析；進一步和傳統的回歸模型預測方法進行比較。研究結果顯示自組性類神經網路的預測效果最佳，且匯率是影響國際觀光客需求的最大因素，而遞延一期的旅客數對旅客需求也有很大的影響。而類神經網路相較於傳統的預測方法更有更高的準確性和穩健性，未來可實際利用類神經網路進行更廣泛的觀光客需求預測應用。

關鍵字

SOM類神經網路 ； BP類神經網路 ； RBF類神經網路 ； 旅客需求 ； 預測模型

並列摘要

Tourism demand is important for policy-making. The study of tourism demand is attracting more and more attention. Research has shown multiple factors may have an impact on tourism demand. However, few studies have been done on applying the multiple factors to forecast tourism demand. The purpose of this thesis is to determine if the model applying artificial neural network would be more suitable for forecasting international tourism demand. The models were established for tourist from Japan to Taiwan. There were three artificial neural network models developed for forecasting tourism demand. The models are including self-organizing map (SOM), radial basis function (RBF), and back-propagation (BP). Rather, factors were specified which, on the basis of previous studies, were applied to the models. The data was collected from 1979 to 2006.Results indicated that SOM models did show the best prediction results and compared to traditional model also show the better result. Exchange rate was the most significant factor for international tourism demand.

並列關鍵字

tourism demand ； artificial neural network ； forecasting model ； SOM ； BP ； RBF

參考文獻

軟科學，21(1): 14-21。

Ao, S. I. (2003) Using fuzzy rules for prediction in tourist industry with uncertainty, Computer Society, 4: 213 -226.

Au, N. and Rob, L. (1999) A neural network model to forecast Japanese demand for travel to Hong Kong, Tourism Management, 20: 89-97.

Carey, G. and Rob, L. (2002) Modeling and forecasting tourism demand for

arrivals with stochastic nonstationary seasonality and intervention, Tourism Management, 23: 499- 510.