Feasibility Analysis of Education in Indonesian Provinces: A Machine Learning Clustering Approach for Regional Classification

Education system feasibility varies significantly across Indonesian provinces due to geographic, socioeconomic, and infrastructural disparities. This study applies unsupervised machine learning clustering algorithms (K-Means, Hierarchical Clustering, Gaussian Mixture Models, and Self-Organizing Maps) to classify Indonesian provinces into homogeneous groups based on education feasibility indicators. Using 39 provinces and special territories, we evaluated clustering quality through internal validation metrics (Silhouette coefficient and Davies-Bouldin index) and inter-algorithm agreement measures (Adjusted Rand Index and Normalized Mutual Information). Results demonstrate that Hierarchical Clustering achieves the best Davies-Bouldin index (0.782), while K-Means and GMM produce identical partitions (ARI = 1.0, NMI = 1.0). Self-Organizing Maps identified nine distinct regional education feasibility profiles, with provincial distributions revealing significant heterogeneity in education conditions. These findings provide a quantitative framework for targeted policy interventions and resource allocation to improve education feasibility across Indonesia’s diverse regions.