Enhancing Hotel Performance Prediction in Oman’s Tourism Industry: Insights from Machine Learning, Feature Analysis, and Predictive Factors

This paper introduces two novel fitness functions adapted to attraction attributes, extending the capabilities of the Linear Genetic Programming for Optimization Decision Tree (LGPDT). A comparative analysis with XGBoost evaluates LGPDT's performance using both traditional and tourism datasets, examining its predictive capacity for hotel performance in Oman. Using extensive experiments and analysis, the study proved the effectiveness of LGPDT in this context, revealing promising results with a mean accuracy of 72.0% and a standard deviation of 7.7%. This underscores LGPDT's robustness and suitability for decision-making in the hospitality industry. Comparison with XGBoost demonstrates LGPDT's slightly higher stability in accuracy, highlighting its potential as a predictive tool. Moreover, the evaluation of new fitness functions reveals that they are computationally efficient while maintaining similar quality standards compared to previously studied fitness functions. These findings underscore LGPDT's efficacy for predicting hotel performance, offering valuable insights for industry stakeholders.