Hendra, Jhidan Daelvin
Unknown Affiliation

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

Analisis Kinerja SVM dan BERT dalam Memprediksi Ketersediaan Layanan Pencegahan Stunting di Indonesia Putra, Bayu Anugerah; Handayani, Fitri; Fatma, Yulia; Hendra, Jhidan Daelvin
JUITA: Jurnal Informatika JUITA Vol. 14 Issue 1, March 2026
Publisher : Department of Informatics Engineering, Universitas Muhammadiyah Purwokerto

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Stunting, defined as being too short for one's age with a Height-for-Age Z-score (HAZ) below –2 SD according to WHO, remains a serious public health problem in Indonesia. This study predicts the availability of stunting prevention services at the village level using machine learning. Data from 25,800 villages were categorized into Complete (9,245), Partial (13,609), and Not Available (2,946), showing class imbalance. Two algorithms were evaluated: Support Vector Machine (SVM) with TF-IDF and SMOTE for class balancing, and Bidirectional Encoder Representations from Transformers (BERT) using IndoBERT with class-weighted loss. Evaluation metrics included accuracy, precision, recall, F1-score, and computation time. Results show BERT achieved 92% accuracy with consistent performance across classes (cross-validation 91.55%, SD 0.0024), effectively capturing contextual meaning in narrative text. SVM reached 83% test accuracy with fast computation (±1 min 42 s) and remained robust for imbalanced data. Both models performed well, but minority-class recognition remains challenging. These findings highlight the complementary strengths of SVM and BERT, providing data-driven insights to support policy decisions and improve targeting of stunting prevention services at the village level.