Classification of diabetic patients with imbalanced class distribution by using a Cost-Sensitive forest algorithm / Ummi Asyiqin Che Muhammad and Muhammad Hasbullah Mohd Razali

In the medical data set, the majority class consist of healthy patients, whereas the minority class consist of a few sick patients. Although many machine learning algorithms have been developed by researchers, the class imbalanced distribution still makes it challenging for classifiers to properly learn and differentiate between the minority and majority classes. This study focused on fitting an imbalanced diabetic data set to a CSForest algorithm. The accuracy of the CSForest was then compared to the RForest. It was found that the accuracy of RForest was 76.70% while the accuracy of the CSForest was 78.72%, indicating that CSForest performs better than the RForest in classifying diabetic patients.