Predicting hotel booking cancellations using tree-based neural network

Area of Science:

• Computer Science
• Artificial Intelligence
• Machine Learning

Background:

• Hotel cancellations disrupt revenue management accuracy.
• Computational advancements enable predictive modeling for risk reduction.
• Existing models require real-world testing and integration into decision support systems.

Purpose of the Study:

• To develop and evaluate a novel computational method for predicting hotel booking cancellations.
• To assess the integration of predictive models into hotel decision support systems.
• To analyze the impact of predictive models on demand-management strategies.

Main Methods:

• Introduction of a tree-based neural network (TNN).
• The TNN combines tree-based learning algorithms with feed-forward neural networks.
• Model testing on two benchmark datasets.

Main Results:

• The TNN model demonstrated significantly improved predictive power.
• Performance was superior compared to traditional tree-based models.
• The TNN outperformed baseline artificial neural networks.

Conclusions:

• Tree-based neural networks show promise for predicting hotel booking cancellations.
• The TNN model offers a viable computational approach for tabular data.
• Further validation in real-world conditions is suggested.