Tourism Demand Forecasting with Tourist Attention: An Ensemble Deep Learning Approach
share
The large amount of tourism-related data presents a series of challenges for tourism demand forecasting, including data deficiencies, multicollinearity and long calculation time. A Bagging-based multivariate ensemble deep learning model, integrating Stacked Autoencoders and KELM (B-SAKE) is proposed to address these challenges in this study. We forecast tourist arrivals arriving in Beijing from four countries adopting historical data on tourist arrivals arriving in Beijing, economic indicators and tourist online behavior variables. The results from the cases of four origin countries suggest that our proposed B-SAKE model outperforms than benchmark models whether in horizontal accuracy, directional accuracy or statistical significance. Both Bagging and Stacked Autoencoder can improve the forecasting performance of the models. Moreover, the forecasting performance of the models is evaluated with consistent results by means of the multi-step-ahead forecasting scheme.
READ FULL TEXT
