Integrating fuzzy C-means clustering and Random Forest for multivariate performance prediction in vocational education

This technical note presents the integration of unsupervised and supervised machine learning methods—Fuzzy C-Means (FCM) clustering and Random Forest regression—for analyzing multivariate determinants of student job performance in vocational education. Using a simulated dataset (N = 300) with seven variables, FCM identified three latent clusters with moderate partition clarity (Partition Coefficient = 0.333; Partition Entropy = 1.099). Random Forest achieved a Mean Squared Error (MSE) of 266.65 and classification accuracy of 38.9% in predicting categorical performance. While predictive power was limited due to simulated and imbalanced data, this framework demonstrates methodological feasibility and highlights key predictors such as confidence, motivation, and supervisor evaluation, serving primarily as a methodological demonstration rather than empirical validation.