This paper presents a study on assessing hotel performance in Oman using a proposed machine-learning technique that integrates evolutionary algorithms in its architecture. The research investigates the issue of population diversity within a previously based genetic programming mechanism. It explores an adaptive population size adjustment technique, proposing enhancements to sustain and enhance population diversity in the genetic programming process. The results illustrate the superiority of this new mechanism over conventional approaches in terms of robustness, accuracy, and stability, particularly evident when applied to tourism datasets. These findings underscore the method’s potential as a practical solution for addressing challenges in machine learning in the tourism sector.