Objective: In recent years, user satisfaction with ecotourism accommodations has become an important determinant of performance and competitiveness in the tourism industry. The rapid growth of online booking platforms, such as Jabama, has intensified competition among accommodation providers and increased the need for data-driven approaches to identify the factors influencing user satisfaction. Although previous studies have highlighted the importance of service quality, cleanliness, and value for money, the application of advanced machine-learning methods to ecotourism accommodations in Iran remains limited. Moreover, the culturally and locally embedded nature of these accommodations calls for analytical approaches that are sensitive to the specific characteristics of rural tourism contexts. This study therefore aims to identify the key determinants of user satisfaction in rural ecotourism accommodations and evaluate the predictive and interpretive capacity of machine-learning algorithms.
Methods: Data from 1,123 active ecotourism accommodations listed on the Jabama platform were collected using the Python programming language and the Selenium web-automation library. User satisfaction scores were considered the dependent variable, while information quality, cleanliness, value for money, reception procedures, hosting quality, and location were treated as independent variables. Five predictive models were employed: linear regression, decision tree, random forest, gradient boosting, and support vector regression (SVR) with a radial basis function (RBF) kernel. Model performance was evaluated using mean squared error (MSE), the coefficient of determination (R²), and five-fold cross-validation. Feature importance was assessed using the random forest model, while the decision tree structure was examined to enhance the interpretability of the findings.
Results: The results showed that the random forest and gradient boosting models achieved the best predictive performance, with R² values exceeding 0.86 and MSE values below 0.012. Feature-importance analysis indicated that information quality was the most influential factor affecting user satisfaction, with an importance coefficient of 0.442, followed by cleanliness and value for money. The decision tree model further supported these findings, identifying information quality as the root node and highlighting the subsequent roles of cleanliness and value for money in the decision-making structure. Spatial analysis also showed that the highest concentration of ecotourism accommodations was found in central provinces, particularly Isfahan, Fars, and Yazd, whereas Alborz, Qom, Semnan, and South Khorasan had the lowest numbers of accommodations.
Conclusions: The findings demonstrate the importance of improving the quality of information and service-related attributes in enhancing user satisfaction with rural ecotourism accommodations. Providing accurate, complete, transparent, and up-to-date information can reduce uncertainty and help tourists make more informed choices, while maintaining high cleanliness standards and an appropriate balance between price and service quality can improve the actual accommodation experience. The proposed data-driven framework provides a practical basis for accommodation managers, online booking platforms, and tourism policymakers to prioritize service improvements, enhance user satisfaction, and strengthen the competitiveness of rural ecotourism destinations.