This study aims to analyze the linear structural relationships and nonlinear predictive contributions associated with tourist satisfaction in the Trusmi Batik Tourism Area. By integrating PLS-SEM, ANN, and SHAP-based XAI, it evaluates how destination attributes function as hygiene or motivator factors. The ultimate goal is to develop a strategic quadrant matrix to guide local governments and destination managers in optimizing craft-based cultural tourism management.
This study employs a multi-stage hybrid design integrating PLS-SEM, an MLP-based artificial neural network (ANN) and explainable artificial intelligence (XAI) using SHAP analysis. PLS-SEM evaluates linear structural relationships based on the proposed theoretical hypotheses. A parsimonious single-hidden-layer ANN captures complex, nonlinear predictive patterns. Finally, SHAP values interpret the model's hidden predictive contributions to construct an actionable strategic quadrant matrix.
The findings suggest that accessibility, resources, and the environment emerged as the primary “Star” motivators, with accessibility showing the strongest structural contribution and predictive importance among the examined constructs (ß = 0.310; |ϕ| = 0.416). “Batik Arts” emerged as the most influential indicator within the dataset. Services occupied the “Foundation” quadrant, whereas Facilities functioned primarily as a baseline hygiene factor and did not exhibit a statistically significant structural relationship with tourist satisfaction (ß = 0.014, p = 0.888). The parsimonious single-hidden-layer ANN demonstrated promising predictive performance in capturing nonlinear patterns within this exploratory dataset.
This study integrates PLS-SEM, ANN, and SHAP-based XAI to examine linear and nonlinear patterns associated with tourist satisfaction within craft-based cultural tourism destinations. The findings suggest prioritizing immersive batik workshops, walkable heritage corridors, digital visitor information, and routine maintenance of basic facilities. The small, geographically focused sample limits generalizability and supports the need for external validation using larger and more diverse datasets.
This study offers a practical reference for tourism operators and local authorities to improve strategic resource allocation. Instead of funding costly luxury hotel expansions, managers could prioritize maintaining hygiene assets, including toilets, seating, and worship facilities. To enhance visitor satisfaction, investments may emphasize “Star” motivators through immersive batik-making workshops and authentic cultural experiences. Local governments may also improve urban micro-mobility by developing walkable sidewalks and flexible motorcycle routes throughout Trusmi's historic heritage corridors. These evidence-based recommendations support sustainable destination management while strengthening visitor experiences, cultural preservation, and long-term tourism resilience.
This study has social implications for sustaining craft-based cultural tourism while strengthening community participation and cultural identity. The findings highlight the importance of accessible public spaces, immersive batik-making experiences, reliable visitor services, and heritage storytelling in supporting visitor engagement. Prioritizing these attributes can help local governments and tourism operators create more inclusive tourism environments that benefit visitors, artisans, and local communities. The framework also supports balanced resource allocation between basic hygiene facilities and culturally distinctive motivators. By strengthening artisan involvement, digital cultural communication, and heritage preservation, the approach can contribute to community resilience, intergenerational knowledge transmission, and sustainable tourism development.
The value of this study lies in its innovative hybrid framework integrating PLS-SEM, MLP-based ANN, and Explainable Artificial Intelligence (XAI) within cultural tourism research. Compared with conventional linear approaches, the framework combines structural relationships, nonlinear predictive patterns, and SHAP-based interpretability to identify destination attributes with different levels of strategic importance.
