Evaluasi Kinerja Model Machine Learning dalam Klasifikasi Penyakit THT: Studi Komparatif Naïve Bayes, SVM, dan Random Forest

Prasetya, Nur Wachid Adi
Wanti, Linda Perdana
Purwanto, Riyadi
Bahroni, Isa
Listyaningrum, Rostika

Infotekmesin
Vol 16 No 2 (2025): Infotekmesin: Juli 2025

Classification of Ear, Nose, and Throat (ENT) diseases is essential to support faster and more accurate diagnosis. However, no prior studies have specifically compared the performance of Naïve Bayes, Support Vector Machine (SVM), and Random Forest algorithms in ENT cases. This study aims to evaluate and compare the three classification models in identifying ENT diseases with or without comorbidities. Medical record data were processed through preprocessing, feature selection using ANOVA, and class balancing with SMOTE. The results showed that SVM outperformed the other models with the highest accuracy (59%), followed by Random Forest (57%), and Naïve Bayes (48%). SVM demonstrated superior performance due to its consistent scores across all evaluation metrics. The study concludes that the choice of classification model significantly impacts the accuracy of ENT disease diagnosis.