Causal Explainable Machine Learning for Identifying Causal Factors in Tourist Decisions at Sustainable Travel Destinations

This research aims to identify the factors that causally influence tourists' decisions at sustainable tourism destinations using a Causal Explainable Machine Learning approach. The research uses a synthetic dataset of 50,000 respondents with 31 variables representing demographic characteristics, travel behavior, environmental awareness, perceptions of destination sustainability, service quality, social influence, digital media exposure, eco-friendly promotions as treatment, perceived value and trust in the destination as mediators, and visit intention as the outcome. The research stages include data pre-processing, descriptive analysis, predictive modeling using Random Forest, propensity score estimation, and Causal Machine Learning using Causal Forest–Double Machine Learning (DML) to estimate the Average Treatment Effect (ATE) and Conditional Average Treatment Effect (CATE). The SHAP (SHapley Additive exPlanations) method is employed to explain feature contributions to tourists' decision predictions and compare predictive importance with factors associated with causal effect heterogeneity. This dataset also provides the original CATE as a reference for evaluating the accuracy of causal effect estimates with results of MAE=0.142061, RMSE=0.171324, and R²=0.712434. The results show that the Causal-XAI approach can distinguish factors that are important for prediction from factors related to the variation in the causal effects of eco-friendly promotions on tourists' decisions. CATE analysis shows that different tourist groups respond differently to sustainable tourism promotion interventions. These findings show that environmental awareness and destination sustainability are important characteristics related to different treatment effects. The proposed approach provides a more transparent, interpretable, and accountable analytical framework to support targeted marketing strategies, improve sustainable tourism experiences, and facilitate data-driven decision-making for sustainable tourism destinations