<![CDATA[Heart disease is a leading cause of death in Indonesia, with a national prevalence of 0.85% in 2025 and cardiovascular-related deaths reaching 651,481 annually. This study aims to implement and compare Random Forest (RF) and Support Vector Machine (SVM) algorithms for heart disease prediction using the UCI Heart Disease dataset (Cleveland subset, 303 instances). Methods include preprocessing (missing value imputation, one-hot encoding, scaling), model training, and evaluation using accuracy, precision, recall, F1-score, and 5-fold cross-validation. Results show RF achieving 86.89% test set accuracy and 89% recall for Presence, while SVM reaches 85.25%. In cross-validation, SVM is more stable with a mean accuracy of 81.84% compared to RF's 80.18%. Feature importance from RF highlights thal_7.0 (0.129), thalach (0.126), and ca (0.110) as key predictors. Conclusion: RF excels in reducing false negatives for early detection, while SVM is more generalizable. The novelty lies in the comprehensive comparison with feature interpretation in the Indonesian health context, supporting AI-based prediction systems for heart disease prevention.]]>
Copyrights © 2025
