Predicting Hotel Booking cancellations

Business Data Analytics | Second Year Group Project

Jayaweera Kankanamge Lihansa Bimandi

Halpewattage Kaawya Yehelini Peiris

Araliya Peiris Malalasekara

Shamla Aqthar

Natania Samuel

Hotel Chain A faces a significant challenge caused by booking cancellations and guest no-shows across its city, airport, and resort properties. These disruptions create uncertainty in occupancy forecasting, reduce revenue, and make it difficult to effectively allocate staff, inventory, and other operational resources. To address this issue, our project developed an AI-powered predictive analytics solution capable of identifying high-risk bookings before the guest’s arrival. The project utilized a dataset of 27,500 historical hotel booking records and followed a comprehensive data analytics and machine learning workflow. This included data profiling, cleaning, feature engineering, exploratory data analysis, and predictive model development. Multiple machine learning models were evaluated, including Logistic Regression, Random Forest, XGBoost, Support Vector Machine, K-Nearest Neighbours, and Artificial Neural Networks, to determine the most effective approach for predicting reservation outcomes. Key booking characteristics such as lead time, length of stay, guest count, weekend stays, meal plans, and booking value were analyzed to identify the factors most strongly associated with cancellations and no-shows. The analysis revealed that approximately 23% of bookings resulted in cancellations or no-shows, with weekend bookings and Full Board reservations showing particularly high risk. Revenue loss associated with these missed stays was estimated at nearly £2 million. Based on these findings, targeted business strategies including risk-based deposits, credit card guarantees, reminder systems, and controlled overbooking opportunities were proposed. The solution enables Hotel Chain A to reduce revenue leakage, improve operational planning, and make more informed business decisions.