A1096
Title: Geopolitical and tourism risks: A comparison of analyses between econometric approaches and machine learning
Authors: Chanon Thongtai - Chulalongkorn University (Thailand) [presenting]
Nuti Sornnil - Chulalongkorn University (Thailand)
Abstract: The aim is to test predictive models and tools between econometric and machine learning methods, as well as to examine whether influence geopolitical risks influence the forecast of tourism demand in Taiwan. Monthly-based data is collected with a total of 264 observations with 8 variables from January 1998 to December 2019, before the COVID pandemic, to provide a shock-free dataset. Various models are used, including: WLS linear regression, k-nearest neighbors, decision tree, random forest, and gradient boosting with metrics as values and $R^2$ and root mean square error (RMSE). The results show that the decision tree model is more effective in forecasting than other types of models. Decision tree is the model with the best indicator values: $R^2 = 0.42$ and RMSE at 0.0899. The econometric analysis suggests that geopolitical risks have a very small, yet negative, effect on the forecast of tourism demand.
