The Study of Forecasting Models for the Visitor Arrivals Demand in Taiwan

• Main Authors: Hong-Bin Chen, 陳鴻彬
• Other Authors: Tzong-Shyuang Chen
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/372g8j

碩士 === 朝陽科技大學 === 休閒事業管理系碩士班 === 94 === In this study, five methods such as single exponential smoothing、H-W exponential smoothing、Fourier series analysis、ARIMA model and Artificial Neural Network were used to establish the Forecasting Models for the Visitor Arrivals in Taiwan. By using the MAPE、RMSPE and Theil’s Inequality Coefficient to access the forecast ability of above methods. The results suggested that H-W exponential smoothing、ARIMA model and Artificial Neural Network are more suitable to forecast the Visitor Arrivals in Taiwan. The Back-propagation Network, BPN, was used in this study to availably increase the precision of fitting by conducting the appropriate trend element in Input layer. Also from the results obtained from sensitivity analysis, it was able to reasonably reduce the number of element of input layer. In accordance with bursting SARS event, the intervention function of sustained unit step function was introduced into the Input layer of Artificial Neural Network in this study. This method was able to precisely simulate both the Outlier during the SARS bursting period and reasonably forecast visitor arrivals in Taiwan during general period. The forcast results indicated about 3 million 500 thousands visitor arrivals in Taiwan for 2006 and 3 million 700 thousands visitor arrivals in Taiwan for 2007. From this trend, it is still having a large gap to achieve the Tourist Doubled Plan as 5 million visitor arrivals in Taiwan for 2008.