Ensembling Methods for Selecting a Splitting Attribute in Decision Trees Learning Algorithms
Table 4: Ensemble efficiency with weighting coefficients
Task number | Classification accuracy for groups: | |||||
All methods | 3 best | 3 average | 3 worst | BAW | Best method | |
1 | 0.419 | 0.64 | 0.548 | 0.409 | 0.413 | 0.691 |
2 | 0.325 | 0.527 | 0.436 | 0.32 | 0.324 | 0.563 |
3 | 0.164 | 0.308 | 0.261 | 0.164 | 0.164 | 0.745 |
4 | 0.864 | 0.884 | 0.864 | 0.828 | 0.853 | 0.883 |
5 | 0.649 | 0.951 | 0.932 | 0.649 | 0.649 | 0.956 |
6 | 0.452 | 0.683 | 0.691 | 0.461 | 0.444 | 0.752 |
7 | 0.889 | 0.951 | 0.897 | 0.843 | 0.942 | 0.957 |
8 | 0.199 | 0.87 | 0.247 | 0.199 | 0.199 | 0.932 |
9 | 0.69 | 0.755 | 0.688 | 0.604 | 0.635 | 0.766 |
10 | 0.541 | 0.826 | 0.543 | 0.541 | 0.541 | 0.831 |
11 | 0.698 | 0.723 | 0.766 | 0.71 | 0.713 | 0.799 |
12 | 0.209 | 0.387 | 0.232 | 0.209 | 0.209 | 0.426 |