Predictive Method for Diabetes using Machine Learning

When the human body lacks sufficient insulin or when body cells become unresponsive to insulin, an excess of glucose remains in the blood. That can lead to serious health problems like renal disease, eyesight loss, and heart disease. Machine learning has the potential to anticipate diabetes. In this paper, the authors have used four classification algorithms: Support vector machine (SVM), K-nearest neighbors (KNN), Random Forest (RF), and Logistic Regression (LR), to predict diabetes and have conducted experiments using the PIMA Indians Diabetes Dataset (PIDD). SMOTE has been utilized to balance the dataset. Class balancing improved performance of classifiers. After implementing class balancing SVM obtained 78% accuracy, KNN obtained 74% accuracy, RF obtained 85% accuracy, and LR obtained 71 % accuracy.