M. Faris Al Hakim, Muthia Nis Tiadah, Subhan
The World Health Organization reported that more than 420 million people around the world are living with diabetes, showing an increase in the prevalence of this condition. It is recognized to be one of the significant contributors to mortality rates and the reduction of life expectancy in the world. Thus, there is a need for effective prevention and early detection measures to combat the global diabetes emergency. This study used person correlation and recursive feature elimination on several models to train the datasets, such as Decision Tree, Logistic Regression, K-Nearest Neighbor (KNN), Extreme Gradient Boost (XGBoost) classifier, and Random Forest, as a classifier. The last step is analyzing model performance using a confusion matrix. This study used a Behavioral Risk Factor Surveillance System (BRFSS) dataset with 330 features. The test results show that the model built can achieve an accuracy value of 91%. © 2025 IEEE.
Department of Computer Science, Universitas Negeri Semarang, Semarang, Indonesia
