CLASSIFICATION OF DENGUE FEVER DISEASE USING A MACHINE LEARNING-BASED RANDOM FOREST ALGORITHM

Abstract

Dengue Hemorrhagic Fever (DHF) is a tropical disease that often results in high morbidity and mortality rates. Early diagnosis of DHF is crucial to mitigate its adverse effects. However, manual diagnostic processes are often inefficient and prone to errors. This study aims to develop a DHF classification model using the Random Forest algorithm, which is expected to assist in the early diagnosis of this disease. The methodology used in this research is CRISP-DM (Cross-Industry Standard Process for Data Mining), which includes the stages of Business Understanding, Data Understanding, Data Preparation, Modeling, Evaluation, and Deployment. Data was obtained from kaggle.com, and during the Data Preparation stage, missing values were removed, categorical features were encoded, data was normalized, and split into training and testing sets. The research results show that the Random Forest model has an accuracy of 88.5%, precision of 88.2%, recall of 65.2%, F1-score of 74.9%, and ROC AUC of 0.810. Feature importance analysis revealed that the Gender_Male and Body_Pain features have the largest contributions in DHF classification. Although the model demonstrated high accuracy and precision, the lower recall value indicates that some positive cases were missed, requiring further improvements. The Random Forest can be used as a tool for early DHF diagnosis, but further adjustments are necessary to enhance its performance. This research provides insights into the contributing factors for DHF diagnosis and the practical application potential of this model in medical decision support systems.