A Machine Learning Approach to Understanding Teacher Engagement in Sustainable Education Systems

Education can be conceptualized as a complex socio-technical system in which teacher engagement functions as a dynamic component supporting system performance and adaptability. The present study examines how science teachers’ perceptions of sustainable education interact with their levels of work engagement, providing empirical insights into system-level relationships relevant to educational sustainability. The study sample consisted of 246 science teachers, and data were collected using the Sustainable Education Scale and the Engaged Teacher Scale. Adopting a systems-informed analytical perspective, the study employs machine learning methods (Random Forest, CART, Extra Trees, and Bagging Regression) to explore non-linear relationships and interaction patterns that may remain obscured in conventional linear analyses. The results indicate that structural factors such as weekly teaching hours and academic qualifications are associated with variations in both sustainable education perceptions and work engagement. Moreover, the findings suggest a reciprocal relationship between sustainability-oriented perceptions and teacher engagement, consistent with feedback dynamics observed in complex educational systems. Rather than proposing a new theoretical framework or algorithm, the study demonstrates the utility of machine learning as a methodological tool for examining system-level interactions and emergent patterns in education, offering empirical insights that may inform sustainability-oriented practices in complex social systems.