Comparison of Support Vector Machine Performance with Oversampling and Outlier Handling in Diabetic Disease Detection Classification

Diabetes mellitus is a disease that attacks chronic metabolism, characterized by the body’s inability to process carbohydrates, fats so glucose levels are high. sixth cause of death in world. Classifying data about diabetes makes it easier predict disease. As technology develops, can be detected using machine learning methods. The method done support vector machine. advantage SVM very effective completing classification, quickly separate each positive and negative point. This study aimed obtain best classification model based on accuracy, sensitivity, precision values detecting adding Synthetic Minority Over-Sampling Technique (SMOTE) handling outliers. SMOTE was applied handle class imbalance. Support Vector Machine (SVM) produce function as dividing line or what called hyperplane matches all input with smallest possible error. studied were indications diabetes, consisting 8-factor variables 1 variable. test results show SVM-SMOTE scenario produces accuracy. produced an accuracy value RBF kernel 88% error 12%, this obtained from division training 90:10. 0.880 sensitivity 0.880. research showed factor more accurate if carried out imbalance (SMOTE), concluded distribution influences scenario.