Forecasting tourist arrivals in Taiwan by using artificial neural network modeling

As one of the important areas in tourism research, tourism demand modeling and forecasting have been attracting an attention of academics and practitioners in decades. Thanks to the perishable nature of products and services in tourism industry, accurate forecasting of tourism demand are crucial especially for operational, tactical, and strategic decision makers. Aside from time-series and regression techniques which have long dominated forecasting models for tourism demand, the purpose of this study is to apply ANN (artificial neural network) models to forecast the tourist arrivals to Taiwan from 8 main inbound countries and demonstrate the potential forecasting capability for ANN models. According to several error measures and statistically hypotheses testified, experiment results show that, no matter in which countries, ANN model outperforms multiple regression, naïve, moving average, and exponential smoothing models