EarlyStage Diabetes Risk Detection Using Comparison of Xgboost, Lightgbm, and Catboost Algorithms

Diabetes Mellitus is a chronic metabolic disorder that elevates blood glucose levels due to insufficient insulin production or insulin resistance. This disease has significant and increasing global impact. Numerous studies have investigated the effects of using classification algorithms to detect and prevent diabetes mellitus. However, existing research faces various challenges in achieving optimal performance and efficient detection time. This study selects the LightGBM, XGBoost, and CatBoost methods for classification. These three algorithms are trained using the “early-stage diabetes risk prediction dataset” and their results are compared to determine which algorithm is best for classifying early-stage diabetes. The testing results indicate that the model trained using the CatBoost algorithm demonstrates superior performance with higher accuracy, precision, recall, F1-Score, and ROC-AUC scores compared to models trained with XGBoost and LightGBM. Additionally, the LightGBM algorithm exhibits faster computational time compared to XGBoost and CatBoost.