MACHINE LEARNING IN HOSPITALITY: INTERPRETABLE FORECASTING OF BOOKING CANCELLATION

Abstract

The hospitality industry is facing significant challenges due to unpredictable booking cancellations, which can negatively impact revenue management, resource allocation, and operational efficiency. With the rapid evolution of online reservation systems and dynamic pricing strategies, traditional forecasting methods often struggle to capture the nuances of customer booking behaviors. This project presents a machine learning framework focused on interpretability for predicting hotel booking cancellations by utilizing historical reservation data and customer characteristics. The system employs a range of machine learning algorithms, such as Decision Trees, Random Forests, Logistic Regression, and Gradient Boosting, to determine the likelihood of cancellations. The dataset includes various attributes, including booking lead time, room types, customer demographics, pricing information, special requests, and prior booking history. To improve predictive performance, data preprocessing techniques such as handling missing values, feature encoding, normalization, and exploratory data analysis are applied. An important feature of this project is the integration of interpretability methods, such as SHAP (Shapley Additive Explanations) and LIME, which provide clear insights into the model's predictions. These tools help hotel managers understand the underlying factors influencing cancellation decisions, thereby increasing trust in machine learning forecasting systems. Evaluations reveal that the proposed framework achieves high accuracy in predictions while maintaining transparency and explainability in its models. The developed system enables hotels to proactively manage cancellation risks, optimize overbooking tactics, enhance customer engagement, and improve their revenue management strategies. By combining predictive analytics with interpretable machine learning, this solution acts as a crucial decision-support tool for modern hospitality operations. It not only boosts forecasting accuracy but also offers actionable insights that support informed business decisions in a competitive environment.