From data to delight: Predictive analytics for optimizing hospitality service ratings

Abstract

As the hospitality industry continues to evolve, machine learning plays a crucial role in improving guest satisfaction and operational efficiency. This research introduces a predictive model specifically designed for homestay businesses, utilizing supervised learning methods-namely decision trees and logistic regression-to anticipate essential rating factors such as cleanliness, communication, accuracy, and check-in experience. The data, collected from a start-up, includes entries with features related to booking methods, customer profiles, and rating scores. Data preprocessing and exploratory analysis are conducted using Python, followed by model training and evaluation to determine guest satisfaction drivers, including factors like stay month and booking channel. The models are validated using accuracy metrics to ensure their reliability and effectiveness in real- world scenarios. Beyond prediction, the study proposes incorporating an AI-based recommendation engine in the future to suggest the most suitable homestay options based on user history and preferences. This combined strategy- rating prediction and personalized recommendations-aims to improve service quality, uncover areas for enhancement, and promote long- term growth in the homestay sector. The flexible design of this framework ensures adaptability across similar businesses, offering scalable, data- driven solutions and providing valuable insights for decision-makers focused on personalized experiences and evidence-based service upgrades.