Table 5 Comparison of the three ensemble methods for both data sets with respect to the macro-F measure
Macro-F measure | ||||||
|---|---|---|---|---|---|---|
Data set | MetaLabeler | SVM Tuned | SVM Vanilla | LLDA | Improve F [13] | MULE Macro |
A | ||||||
✓ | ✓ | 0.53390 △ | 0.54820 | |||
✓ | ✓ | 0.53221 △ | 0.54921 | |||
✓ | ✓ | 0.42563 △ | 0.47918 | |||
✓ | ✓ | ✓ | 0.49437 △ | 0.51099 | ||
✓ | ✓ | ✓ | ✓ | 0.52487 △ | 0.54847 | |
B | ||||||
✓ | ✓ | 0.42573 △ | 0.43342 | |||
✓ | ✓ | 0.42429 △ | 0.43212 | |||
✓ | ✓ | 0.37556 | 0.37335 | |||
✓ | ✓ | ✓ | 0.38149 | 0.38058 | ||
✓ | ✓ | ✓ | ✓ | 0.42240 △ | 0.43324 | |