Penerapan Recursive Feature Elimination (RFE) pada Tree-Based Classifier untuk Identifikasi Risiko Diabetes

Diabetes mellitus is a common chronic disease with significant global impact. Early identification of individuals at high risk of developing diabetes is critical for the prevention and management of the disease. This study explores the use of Recursive Feature Elimination (RFE) in decision tree-based classifiers to improve the accuracy of diabetes risk prediction. The Pima Indians Diabetes Database (PIDD) dataset was used as the database, and algorithms such as Decision Tree, Random Forest, Gradient Boosting, and Xtreme Gradient Boosting were tested. The results showed that the application of RFE improved the model accuracy, with Random Forest and Gradient Boosting achieving the highest accuracy of 77.27%. RFE also successfully identified the most relevant features, reduced the risk of overfitting, and improved model interpretability. This study provides a strong foundation for the development of more effective predictive tools in diabetes management and prevention. Future studies are recommended to test the generalizability of this approach to a wider dataset and in various clinical contexts.