<![CDATA[Purpose – This study aims to develop and evaluate a machine learning-based classification framework for predicting teacher professional development outcomes in North Sulawesi and to compare the predictive performance of several supervised learning algorithms Methods – A quantitative cross-sectional predictive analytics design was employed involving 520 teachers from primary, lower secondary, and upper secondary schools. Data were collected through validated questionnaires and training evaluation instruments measuring demographic, motivational, institutional, and instructional variables grounded in the Professional Learning Community (PLC) and Technological Pedagogical Content Knowledge (TPACK) frameworks. Five machine learning algorithms, namely Support Vector Machine (SVM), Random Forest (RF), k-Nearest Neighbors (kNN), Naïve Bayes (NB), and Decision Tree (DT), were implemented and evaluated using 20-fold stratified crossFindings – The results revealed that SVM achieved the best performance with an AUC of 93.6%, accuracy of 86.9%, F1-score of 86.9%, and MCC of 73.5%, outperforming RF, NB, DT, and kNN. The findings indicate that teacher development outcomes are influenced by complex interactions among motivational, organizational, and technological factors. Institutional support, collaborative school culture, and technological readiness emerged as critical determinants of successful professional development. Research implications – The cross-sectional design and reliance on questionnaire-based measures limit causal interpretation and generalizability beyond the study context. Originality –This study extends educational data mining research beyond student-centered analytics by integrating PLC and TPACK perspectives into a machine learning framework for teacher professional development evaluation in a developing-country context, providing evidence-based insights for educational policy and intervention planning.]]>
