Comparative Evaluation of Naïve Bayes and Support Vector Machine for Human Development Index Classification

Abstract

The Human Development Index (HDI) is one of the most widely used indicators for evaluating the quality of human development through health, education, and living standards. Accurate HDI classification is essential for supporting evidence-based public policy and regional development planning. Recent advances in machine learning provide opportunities to improve classification performance compared with conventional statistical approaches. This study aims to compare the performance of the Naïve Bayes and Support Vector Machine (SVM) algorithms for classifying the Human Development Index across Indonesian provinces and to identify the more effective classification model. The study employed a quantitative experimental approach using the 2024 Human Development Index dataset published by Statistics Indonesia (BPS), covering all 38 provinces. Four predictor variables were used: life expectancy at birth, expected years of schooling, mean years of schooling, and adjusted expenditure per capita. The dataset was preprocessed before being divided into training and testing sets using an 80:20 ratio. Both algorithms were implemented using Python's Scikit-learn library and evaluated through confusion matrix analysis, accuracy, precision, recall, and F1-score. Experimental results indicate that both algorithms successfully classified provincial HDI categories. However, SVM consistently outperformed Naïve Bayes, achieving an accuracy of 100% on the held-out test set (5-fold cross-validated mean accuracy of 86.4%), compared with 87.5% (cross-validated mean of 59.6%) obtained by Naïve Bayes. SVM also produced higher precision, recall, and F1-score, indicating stronger classification capability for multidimensional socioeconomic data. The findings demonstrate that Support Vector Machine provides superior performance for HDI classification compared with Naïve Bayes. This study contributes empirical evidence regarding the application of machine learning techniques in socioeconomic indicator classification and offers practical insights for decision support in regional development policy.