Using XGBoost to Evaluate the Impact of Tourism Destination Marketing Campaigns on Tourists’ Psychological Motivations and Behavioral Intentions

With the rapid development of mobile communication networks and digital information dissemination technologies, tourism destination marketing increasingly relies on wireless media platforms to influence tourists’ psychological responses and behavioral decisions. To reveal the nonlinear mechanisms underlying this process, this study proposes a multi-level attribution framework integrating Extreme Gradient Boosting (XGBoost) and SHAP analysis. Using 4,826 field survey samples combined with real marketing logs, the framework models the relationships among marketing activities, psychological motivations, and behavioral intentions while quantifying feature contributions and mediation effects. The results demonstrate that short-video exposure exerts the strongest influence on recommendation intention among first-time out-of-province tourists, whereas festival activities more effectively promote revisit intention among repeat visitors. Heterogeneity analysis further reveals age-dependent differences in motivational pathways, with social identity dominating younger groups and escapism becoming more influential among older tourists. The proposed interpretable machine learning framework provides robust support for precision marketing and behavioral prediction under data-rich environments. Moreover, its methodology offers potential value for communication-assisted intelligent service systems and multimedia information dissemination scenarios, where reliable digital content transmission and user behavior modeling are essential for optimizing personalized interaction strategies.