Klasifikasi Indeks Pembangunan Manusia Wilayah 3T Di Indonesia Menggunakan Metode Support Vector Machine

Abstract

The Human Development Index (HDI) is an important indicator used to measure the quality of human development through the dimensions of health, education, and decent living standards. The imbalance in HDI category distribution in Indonesia’s 3T (Disadvantaged, Frontier, and Outermost) regions may affect the performance of classification models, especially in recognizing minority classes. This study aims to apply the Support Vector Machine (SVM) method to classify the HDI of 62 regencies/cities in 3T regions using five predictor variables, namely life expectancy, expected years of schooling, mean years of schooling, expenditure per capita, and percentage of poor population. Data preprocessing was carried out using Min-Max Scaling normalization. To address class imbalance, the SVM model was implemented with class weight balanced. The data were divided into three proportions: 60:40, 70:30, and 80:20. The best parameter selection was performed using GridSearchCV with linear, RBF, and polynomial kernels. The results showed that the best model was obtained using the RBF kernel with a data proportion of 80:20, producing an accuracy of 92%, precision of 89%, recall of 96%, and F1-score of 91%. The confusion matrix indicated that only one data point was misclassified, namely North Lombok Regency, which was originally in the medium HDI category but predicted as high HDI. Overall, the SVM method with the RBF kernel was able to classify HDI in Indonesia’s 3T regions effectively and accurately.