<b><i>COMPARATIVE ANALYSIS OF DECISION TREE AND RANDOM FOREST ALGORITHMS FOR PREDICTING DIABETES MELLITUS</i></b>

Diabetes mellitus (DM) is a chronic disease with an increasing number of sufferers and a risk of serious complications. Early detection is very important to prevent these risks. This study uses a public dataset from Kaggle to compare the performance of Decision Tree and Random Forest algorithms in predicting diabetes status. The dataset includes demographic and medical information such as age, hypertension, cardiac history, BMI, HbA1c, and blood glucose levels. Unbalanced data was handled using the SMOTE method, and then tested with 80:20, 70:30, and 60:40 data sharing schemes. The evaluation results showed that Random Forest excelled in all schemes, with the best performance in the 60:40 scheme (96.02% accuracy, 76.13% F1-score). This research shows that Random Forest is effective to support machine learning-based diabetes early detection system.