A HYBRID SARIMA-BILSTM-GRADIENT BOOSTING ALGORITHM IN CRUISE DEMAND PREDICTION AND EVALUATING THE SUCCESS OF CRUISE TOURISM

The purpose of this paper is to use demand prediction algorithms with a strategic perspective for the management and success of cruise tourism.Tourism demand series are frequently noisy, conditionally nonstationary, and, in some situations, deterministically chaotic due to the complexity and constantly evolving nature of the tourism sector.This study presents a new methodology for cruise tourism that uses a Seasonal Autoregressive Integrated Moving Average (SARIMA), Bidirectional Long Short-Term Memory (BiLSTM), Extreme Gradient Boosting (XGBoost), Categorical Boosting (CatBoost), Adaptive Boosting (AdaBoost) and ensemble models to predict the number of cruise-type passenger ships, arriving passengers, and departing passengers.The results highlight how ensemble models can be integrated to produce more accurate and useful prediction in the ever-changing tourism sector.The prediction results can provide actionable insights for policymakers and destination managers in preparing for future growth in cruise tourism.