Tourism demand forecasting using short video information

Abstract

Based on short video information, this study extracted two explanatory variables, popularity and publicity, to empirically forecast weekly tourism demand for a destination (Macao) and a tourist attraction (Mount Siguniang, China). Results indicated that 1) models integrating the popularity or publicity of short videos outperform models without these attributes in tourism demand forecasting; 2) compared with popularity, models featuring publicity from short videos can generate more accurate forecasts; 3) models combining publicity and popularity do not necessarily exceed the performance of models including only publicity; and 4) when models account for search queries as well as publicity, search queries help improve forecasting accuracy for tourist attractions (this positive impact does not apply to destinations).