Research Article
[Retracted] The Use of Hellinger Distance Undersampling Model to Improve the Classification of Disease Class in Imbalanced Medical Datasets
Table 9
The average results using classifiers (KNN, SVM, and DT) with models (baseline, Tml, RUS, ENN, and HDUS) on four datasets.
| | | Baseline | Tml | RUS | ENN | HDUS |
| KNN | Sensitivity (%) | 30.70 | 36.55 | 58.37 | 47.02 | 62.28 | Specificity (%) | 88.54 | 81.18 | 65.12 | 74.34 | 65.39 | Precision (%) | 37.54 | 33.47 | 38.60 | 37.24 | 36.88 | F1_m (%) | 33.59 | 34.92 | 46.06 | 40.82 | 45.76 | Bacc (%) | 59.37 | 58.87 | 61.77 | 60.68 | 63.84 |
| SVM | Sensitivity (%) | 7.03 | 33.61 | 67.63 | 42.57 | 73.38 | Specificity (%) | 96.66 | 88.71 | 60.01 | 80.62 | 60.36 | Precision (%) | 20.06 | 37.86 | 38.49 | 38.73 | 39.10 | F1_m (%) | 10.34 | 35.19 | 48.03 | 39.98 | 49.85 | Bacc (%) | 51.85 | 61.16 | 63.94 | 61.69 | 66.87 |
| DT | Sensitivity (%) | 39.82 | 41.91 | 55.25 | 55.04 | 82.91 | Specificity (%) | 75.34 | 70.51 | 54.02 | 65.89 | 57.44 | Precision (%) | 36.60 | 33.61 | 30.10 | 36.66 | 41.65 | F1_m (%) | 37.86 | 36.23 | 38.60 | 43.59 | 54.79 | Bacc (%) | 57.58 | 56.21 | 54.64 | 60.47 | 70.17 |
|
|