Penerapan Metode Naive Bayes dan Support Vector Machine (SVM) dalam Mengklasifikasikan Status Gizi Balita Stunting di Puskesmas Sumbersari Kabupaten Jember

Abstract

Stunting is a chronic nutritional problem that has serious impacts on the growth and development of toddlers. This study aims to compare the performance of the Naïve Bayes and Support Vector Machine (SVM) algorithms in classifying the nutritional status of at-risk toddlers into two categories: stunting and not stunting.. The research data were obtained from Puskesmas Sumbersari, Jember Regency in 2023, consisting of 361 records, and processed using the Synthetic Minority Oversampling Technique (SMOTE) to address class imbalance, increasing the dataset to 612 records. The study used the following attributes: gender, age (in days), weight (kg), and height (cm). All data underwent preprocessing stages, including cleaning, normalization, label encoding, and balancing. The implementation compared both algorithms using three data splitting ratios: 70:30, 80:20, and 90:10, with evaluations based on the confusion matrix, calculating accuracy, precision, recall, and F1-score. The evaluation results showed that the SVM with RBF kernel yielded the best performance with 94% accuracy, 95.5% precision, 91.5% recall, and 93% F1-score at the 90:10 ratio. In comparison, the Naïve Bayes algorithm achieved 87% accuracy, 87.5% precision, 85% recall, and 86% F1- score. SVM's advantage lies in its ability to handle complex interactions between variables and nonlinear patterns, whereas Naïve Bayes assumes feature independence. These findings indicate that SVM is more reliable and accurate in classifying toddler nutritional status and is more suitable for use as a decision support system in early detection and intervention programs for stunting cases, particularly at the primary healthcare (puskesmas) level.