Research Article

AWSMOTE: An SVM-Based Adaptive Weighted SMOTE for Class-Imbalance Learning

Table 8

G-mean with different oversampling methods from real datasets.

IdNO-RSSMOTEBLSMOTEADASYNDBSMOTEAWSMOTE

10.96090.97000.96230.96260.95790.9717
20.82590.91170.88960.90830.91310.9173
30.18700.9904——0.99530.99480.9972
40.95260.9640——0.96090.97000.9709
50.66860.91500.93700.94750.91110.9519
60.89230.89700.92040.92280.89710.9270
70.70690.99270.98520.99300.98750.9934
80.26350.83930.85870.89850.81180.9001
90.87950.93440.93800.93050.87420.9482
100.94740.95310.97280.97280.97490.9763
110.88650.95500.95450.95220.94890.9556
120.86470.9930——0.98350.99430.9959
130.04040.9871——0.98590.98700.9874
1400.9962——0.99520.98960.9978
150.02820.9765——0.99960.99881
160.00570.9939——0.99370.99390.9972
170.04240.9987——0.99850.98801
180.84590.9878——0.98630.97760.9935
190.86290.9796——0.99040.98740.9941
200.9523——————0.99260.9959
2100.9798——0.97740.99060.9875
2200.9931——0.99410.99220.9946