Research Article
Voting Classification-Based Diabetes Mellitus Prediction Using Hypertuned Machine-Learning Techniques
| Classifier/algorithm | Technique | Accuracy % | ROC % |
| Logistic regression | Default Tomek- undersampling Smote- oversampling | 77.7 77.0 76.7 | 68.3 69.7 76.5 | SVM | Default Tomek- undersampling Smote- oversampling | 79.2 77.7 7.2 | 71.2 70.2 70.3 | KNN | Default Tomek- undersampling Smote- oversampling | 74.5 78.2 77.5 | 66.5 73.3 77.9 | Gradient boost | Default Tomek- undersampling Smote- oversampling | 79.4 77.0 77.9 | 71.6 72.0 78.0 | Naive Bayes | Default Tomek- undersampling Smote- oversampling | 79.7 75.5 73.9 | 73.1 72.0 73.8 | Random Forests | Default Tomek- undersampling Smote- oversampling | 79.7 75.9 80.7 | 73.5 69.8 80.6 | Voting classifier (using three best models as an input) | Default Tomek- undersampling Smote- oversampling | 81.7 77.7 81.5 | 81.6 76.5 81.5 |
|
|