Evaluating Community Training Effectiveness for Blue Economy and Circular Economy Implementation: A Hybrid SEM–Machine Learning Approach in the Citarum River Basin

Community capacity building in the Citarum River Basin is critical for sustainable environmental management through the principles of the Circular Economy (CE) and the Blue Economy (BE). This study developed an integrated modeling approach combining Structural Equation Modeling (SEM) and Machine Learning (ML) to analyze the effectiveness of CE- and BE-based training programs designed to strengthen community capacity in the Citarum River Basin. The factors examined include individual characteristics, teaching quality, and organizational and environmental support, with participant commitment serving as a mediating variable in influencing training effectiveness and sustainable environmental management outcomes. In this framework, SEM was employed to validate theoretical constructs and test causal relationships among latent variables, while ML techniques, specifically Random Forests and Artificial Neural Networks (ANNs), were incorporated to enhance predictive capabilities beyond what theory-driven models alone can achieve. The results demonstrate that all exogenous variables significantly influence training performance, both directly and indirectly, with organizational and environmental support as the most dominant factor, and that the integrated SEM-ML model outperforms the standalone SEM model. The integrated SEM-ML model yielded lower prediction error rates and higher explanatory power, with SEM-ANN delivering the best overall performance. These findings underscore the value of integrating theory-based and data-driven approaches in capacity-building research, effectively addressing the trade-off between model interpretability and predictive accuracy, and providing actionable insights for designing more impactful community training programs to support sustainable management of the Citarum River Basin.