Predictive Modeling of Tourist Satisfaction Based on Service Marketing Mix Elements Using Machine Learning Techniques

This study examines the impact of the service marketing mix on tourist satisfaction and loyalty, focusing on Cox's Bazar, Bangladesh. Utilizing data collected from 500 respondents and analyzed through advanced statistical and machine learning techniques, the study provides key insights into the relationships between marketing mix elements and tourist satisfaction. The reliability of the constructs was assessed using Cronbach's alpha, all of which exceeded the acceptable threshold of 0.70, indicating strong internal consistency. Multicollinearity issues among predictors were resolved by aggregating closely related variables, reducing the variance inflation factor (VIF) to below 1.05. Principal component analysis (PCA) demonstrated that the first two components accounted for 97.88% of the variance, emphasizing the compactness of the data. Predictive modeling revealed that XGBoost outperformed other models with the lowest mean squared error (MSE = 0.10), root mean squared error (RMSE = 0.33), and mean absolute error (MAE = 0.25), alongside the highest R-squared value of 0.74. Feature importance analysis highlighted that the combined variable price_place_aggregated contributed most significantly (68.20%) to the model's predictions, followed by promotion and process. Cross-validation confirmed the robustness of the XGBoost model, with a cross-validated MSE of 0.1273 ± 0.0170. The findings underscore the critical role of pricing strategies and location in enhancing tourist satisfaction and loyalty. This research validates the stability and reliability of the model by integrating sensitivity analysis and learning curve evaluations. These findings offer pragmatic recommendations for policymakers and tourism stakeholders in Cox's Bazar to enhance their marketing strategies and enhance the entire tourist experience.