A decomposition-ensemble approach for tourism forecasting.

Area of Science:

• Tourism and Hospitality Management
• Data Science and Forecasting
• Artificial Intelligence in Economics

Background:

• Recent irregular events have caused significant volatility in tourism demand, particularly in regions like Hong Kong.
• Accurate tourism demand forecasting is crucial for effective industry management and strategic planning.
• Existing forecasting models struggle with the complexity and volatility inherent in tourism time series data.

Purpose of the Study:

• To develop and evaluate a novel decomposition-ensemble approach for enhancing tourism demand forecasting accuracy.
• To investigate the application of data characteristic analysis within a decomposition-ensemble framework.
• To demonstrate the model's effectiveness using Hong Kong tourism demand as a case study.

Main Methods:

• A decomposition-ensemble approach integrating complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and data characteristic analysis.
• Utilizing Elman's neural network model for the ensemble forecasting component.
• Empirical analysis of Hong Kong tourism demand data to validate the proposed methodology.

Main Results:

• The proposed decomposition-ensemble model significantly outperforms traditional models in both point and interval forecasts.
• The model demonstrates superior predictive accuracy across various prediction horizons.
• Data characteristic analysis proved effective in enhancing the decomposition-ensemble approach for complex time series.

Conclusions:

• The developed decomposition-ensemble approach offers a robust and effective solution for tourism demand forecasting.
• The methodology is particularly well-suited for handling complex and volatile tourism time series.
• This research provides valuable insights for improving predictive accuracy in the tourism industry.