Analisis Klasifikasi Tingkat Infeksi Saluran Pernapasan Akut di Puskesmas Setiabudi dengan Metode Random Forest

Abstract

Acute Respiratory Infection (ARI) is an infectious disease affecting the respiratory tract that requires appropriate treatment according to its severity. This study aims to implement the Random Forest algorithm to classify ARI severity based on patients' clinical symptoms at Setiabudi Community Health Center and evaluate its performance using K-Fold Cross Validation. The dataset consisted of 3,736 secondary medical records from 2025 with 14 variables, including 13 clinical features (age, respiratory rate, cough, runny nose, fever, sore throat, nausea and vomiting, diarrhea, contact with ARI patients, body aches, lump, history of asthma, and test results) and one target variable representing ARI severity (Mild ARI and Severe ARI). The research process included data preprocessing, data encoding, feature selection, Random Forest model development, and performance evaluation using Accuracy, Precision, Recall, F1-Score, Classification Report, and Confusion Matrix. Precision, Recall, and F1-Score were calculated using the weighted average method due to class imbalance. The model achieved an Accuracy of 95.45%, Precision of 93.57%, Recall of 95.45%, and F1-Score of 94.08%, exceeding the research target of 80% accuracy. These results indicate that Random Forest combined with K-Fold Cross Validation is effective for classifying ARI severity.