Purpose: This study aimed to analyze guest sentiment from reviews on Booking.com to identify insights that may help reduce room reservation cancellations.

Methodology/approach: A quantitative approach was used, with data collected through web scraping using Python from customer reviews on the Booking. com website. A total of 433 reviews were analyzed using the Naïve Bayes classification method for sentiment analysis.

Results/findings: The analysis revealed that 362 reviews (83.6%) contained positive sentiments, indicating high guest satisfaction, particularly with staff service, room quality, and facilities such as the pool and breakfast. Meanwhile, 71 reviews (16.4%) expressed negative sentiments, mainly focusing on room quality and overall hotel experience. The Naïve Bayes model achieved a classification accuracy of 91%, with a high F1-score of 95% for positive sentiments but only 31% for negative sentiments, highlighting data imbalance. Based on these findings, hotel management is advised to pay more attention to key aspects such as “staff,” “room,” “pool,” and “breakfast” to enhance guest satisfaction and minimize reservation cancellations.

Conclusion: Most reviews reflected positive sentiments, indicating a high level of satisfaction. However, negative reviews, although fewer, must be further evaluated to improve service quality, especially given the classification model’s lower performance on negative sentiments.

Limitations: This study is limited to Booking.com reviews for The Westin Resort & Spa Ubud, based on 433 entries.

Contribution: This study provides a sentiment analysis approach to help hotel management better understand customer feedback and develop strategies to reduce cancellation rates.