Table 4 Average F1 scores on different N values. The values in brackets represent the standard deviation (Std) in 5-fold cross validation. The proposed SASMOTE algorithms perform better on the average F1 score with respect to the value of N = 1, 2, 3, 5 while B-SMOTE performs better with respect to the value of N = 4
Model | \(\boldsymbol{N = 1}\) (Std) | \(\boldsymbol{N = 2}\) (Std) | \(\boldsymbol{N = 3}\) (Std) | \(\boldsymbol{N = 4}\) (Std) | \(\boldsymbol{N = 5}\) (Std) |
|---|---|---|---|---|---|
SASMOTE | 47.04 (7.22) | 51.32 (7.70) | 51.88 (6.60) | 51.13 (8.51) | 51.16 (6.32) |
SASMOTE w/o visible | 48.97 (7.90) | 49.60 (10.14) | 48.04 (8.48) | 50.15 (8.55) | 49.72 (9.67) |
SASMOTE w/o inspection | 48.85 (4.60) | 48.30 (6.90) | 53.34 (6.92) | 46.56 (8.90) | 50.26 (11.94) |
B-SMOTE | 44.76 (7.85) | 49.37 (10.60) | 49.31 (7.76) | 51.17 (10.53) | 47.80 (8.12) |
SMOTE | 45.59 (8.90) | 46.31 (9.71) | 47.72 (11.81) | 50.03 (7.92) | 46.07 (11.19) |