Fig 1.
Depicts diagram of conventional SMOTE algorithm.
Fig 2.
Depicts the two dataset described in Table 1.
DTS1 above shows classes with more pockets of bunching together (Clustered) whereby the DTS2 is a more sporadic class dataset (Sparser).
Table 1.
Shows parameters of the datasets used in this study.
Fig 3.
Schematic diagram illustrating the imbalanced learning workflow.
Table 2.
Top 10 performing oversamplers for DTS1 versus baseline (no oversampling and SMOTE) averaged across four classifiers.
Table 3.
Top 10 performing oversamplers for DTS2 versus baseline (no oversampling and SMOTE) averaged across four classifiers.
Table 4.
Shows the average and top performing AS over all oversamplers for the four different classifier types in DTS1 and DTS2.
Table 5.
Shows the top performers ranked by AS scores over the four columns reporting the four classifier techniques used.
Table 6.
Table comparing operating principles over DTS1 and DTS2 based upon oversamplers categorized in S3 Table.