Due to high fluctuations in booking behaviors, traditional forecasting models struggle to capture underlying trends. The researchers employed clustering to group similar booking patterns, followed by ensemble modeling to improve forecast accuracy. Various forecasting models, including Historical Average, Ordinary Least Squares, Weighted Least Squares, and ensemble models, were tested using Sunweb's booking data.

The results reveal that ensemble models, especially those utilizing clustering, achieved the lowest forecasting errors, outperforming general forecasting models and a benchmark model. The Weighted Average Model consistently showed strong performance across multiple evaluation metrics, providing Sunweb with valuable insights for more effective resource allocation, dynamic pricing strategies, and improved forecasting capabilities.