Table 1.
Candidate estimators used to estimate . See the SuperLearner package documentation for more details (Polley vanderLaan12). SL.earth only used for V = (W1,…,W4).
| Loss function | Method | R function |
|---|---|---|
| Squared error | Bayesian GLM | SL.bayesglm |
| Generalized additive model | SL.gam | |
| Generalized linear model | SL.glm | |
| Generalized linear model, interactions | SL.glm.interaction | |
| Multivariate adaptive regression splines | SL.earth | |
| Sample mean | SL.mean | |
| Neural network | SL.nnet | |
| Stepwise regression | SL.step | |
| Forward stepwise regression | SL.step.forward | |
| Stepwise regression, interactions | SL.step.interaction | |
| Weighted log | Generalized additive model | SL.gam |
| Generalized linear model | SL.glm | |
| Generalized linear model, interactions | SL.glm.interaction | |
| Neural network | SL.nnet | |
| Recursive partitioning | SL.rpart |