Table 6.:
Non-zero weights of individual algorithms in Super Learners 1 and 2 across all three data sets.
| Data Set | Algorithms Selected for SL1 | Weight |
|---|---|---|
| NOAC | SL.caret.bayesglm_All | 0.30 |
| SL.caret.C5.0_All | 0.11 | |
| SL.caret.C5.0Tree_ll | 0.11 | |
| SL.caret.gbm_All | 0.39 | |
| SL.caret.glm_All | 0.01 | |
| SL.caret.pda2_All | 0.07 | |
| SL.caret.plr_ll | 0.01 | |
| NSAID | SL.caret.C5.0_All | 0.06 |
| SL.caret.C5.0Rules_All | 0.01 | |
| SL.caret.C5.0Tree_All | 0.06 | |
| SL.caret.ctree2_All | 0.01 | |
| SL.caret.gbm_All | 0.52 | |
| SL.caret.glm_All | 0.35 | |
| VYTORIN | SL.caret.gbm_All | 0.93 |
| SL.caret.multinom_All | 0.07 | |
| Data Set | Algorithms Selected for SL2 | Weight |
| NOAC | SL.caret.C5.0_screen.baseline | 0.03 |
| SL.caret.C5.0Tree_screen.baseline | 0.03 | |
| SL.caret.earth_screen.baseline | 0.05 | |
| SL.caret.gcvEarth_screen.baseline | 0.05 | |
| SL.caret.pda2_screen.baseline | 0.02 | |
| SL.caret.rpart_screen.baseline | 0.04 | |
| SL.caret.rpartCost_screen.baseline | 0.04 | |
| SL.caret.sddaLDA_screen.baseline | 0.03 | |
| SL.caret.sddaQDA_screen.baseline | 0.03 | |
| SL.hdps.100_All | 0.00 | |
| SL.hdps.350_All | 0.48 | |
| SL.hdps.500_All | 0.19 | |
| NSAID | SL.caret.gbm_screen.baseline | 0.24 |
| SL.caret.sddaLDA_screen.baseline | 0.03 | |
| SL.caret.sddaQDA_screen.baseline | 0.03 | |
| SL.hdps.100_All | 0.25 | |
| SL.hdps.200_All | 0.21 | |
| SL.hdps.500_All | 0.01 | |
| SL.hdps.1000_All | 0.23 | |
| VYTORIN | SL.caret.C5.0Rules_screen.baseline | 0.01 |
| SL.caret.gbm_screen.baseline | 0.71 | |
| SL.hdps.350_All | 0.07 | |
| SL.hdps.750_All | 0.04 | |
| SL.hdps.1000_All | 0.17 |