This study aims to predict tourist arrivals in Sri Lanka by employing advanced machine learning (ML) models and incorporating social media data from platforms like TripAdvisor and Google Trends. The study compared three ML models (Support Vector Regression, Random Forest, and Artificial Neural Network) with a traditional Seasonal Auto-Regressive Integrated Moving Average (SARIMA) model, using historical tourist arrival data as features. The results indicate that ML models generally outperform SARIMA, especially during the turbulent period of 2019-2021 in Sri Lanka. When social media data is integrated, the Random Forest model stands out as the most effective, while the Support Vector Regression model shows less improvement. Although the Artificial Neural Network model doesn't outperform others with social media data, it proves adept at capturing data trends. This study is pioneering in its exploration of ML models and social media integration for predicting tourist arrivals in Sri Lanka, offering valuable insights for industry stakeholders to make informed decisions in the dynamic tourism sector.

I. U. Hewapathirana

Journal of Tourism Futures 

Abstract:- https://doi.org/10.1108/JTF-06-2023-0149