<![CDATA[Diabetes mellitus is a chronic disease with an increasing prevalence worldwide, requiring early detection to support faster and more accurate disease management. This study aims to compare the performance of several machine learning algorithms for diabetes classification using feature selection and hyperparameter tuning. The dataset used was the Pima Indians Diabetes Dataset obtained from the Kaggle platform. The research consisted of data preprocessing, feature selection using SelectKBest, training and testing data splitting, hyperparameter tuning using GridSearchCV, and model evaluation using accuracy, precision, recall, F1-score, ROC-AUC, and cross validation. The evaluated algorithms included Logistic Regression, Support Vector Machine (SVM), Random Forest, K-Nearest Neighbor (KNN), and Naive Bayes. The results showed that the KNN algorithm achieved the best performance with an accuracy of 74.02%, precision of 63.46%, recall of 61.11%, F1-score of 62.26%, and ROC-AUC of 79.60%. The findings indicate that integrating data preprocessing, feature selection, hyperparameter tuning, and cross validation provides a more comprehensive evaluation process for machine learning models in diabetes classification.]]>
Copyrights © 2026
