<![CDATA[Background: Hotel booking cancellation may cause financial losses and reduce hotel operational effectiveness.Objective: This study aims to analyze the factors influencing hotel reservation cancellations and develop a cancellationprediction model to support decision-making in the hospitality industry.Methods: The research method used is CRISP-DM (Cross Industry Standard Process for Data Mining) with a Big Dataand Machine Learning approach based on PySpark. The dataset used is the Hotel Booking Demand Dataset consisting of 119,390 reservation records. Data processing stages include data cleaning, duplicate data removal, categorical data encoding, feature assembling, and model development using Spark MLlib. The algorithms used in this study are Logistic Regression as a baseline model and Random Forest Classifier as the main prediction model.Result: The results show that Random Forest achieved the best performance with an Accuracy of 78.42%, an F1 Scoreof 76.83%, and a ROC AUC of 80.04%. Based on feature importance analysis, the most influential factors affectingreservation cancellation are lead time, market segment, and total special requests.Conclusion: The developed model can be used as a basis for implementing reservation risk scoring, enabling hotels to identify high-risk bookings and formulate more effective cancellation mitigation strategies]]>
Copyrights © 2026
