Table 2.
balancing strategy | classifier | MCC cv | MCC val |
---|---|---|---|
adasyn | RF | 0.63 (0.60, 0.66) | 0.12 |
oversampled_all | RF | 0.69 (0.65, 0.71) | -0.13 |
oversampled_minority | RF | 0.69 (0.65, 0.71) | -0.13 |
smote | RF | 0.63 (0.60, 0.66) | 0.02 |
smote_svm | RF | 0.61 (0.59, 0.65) | -0.09 |
smote_borderline1 | RF | 0.61 (0.58, 0.64) | -0.04 |
smote_borderline2 | RF | 0.59 (0.55, 0.63) | -0.07 |
adasyn | NBM2 | 0.07 (0.03, 0.10) | 0.02 |
oversampled_all | NBM2 | 0.24 (0.19, 0.29) | -0.02 |
oversampled_minority | NBM2 | 0.23 (0.19, 0.28) | 0.07 |
smote | NBM2 | 0.20 (0.15, 0.25) | -0.2 |
smote_svm | NBM2 | 0.24 (0.20, 0.29) | 0.1 |
smote_borderline1 | NBM2 | 0.23 (0.19, 0.29) | -0.11 |
smote_borderline2 | NBM2 | 0.11 (0.06, 0.16) | -0.01 |
Boldface indicates the best performance of RF or NBM2 models either in cross validation or in validation