Using ensemble to train and test
We can fit more than one model in Super Learner and it will tell which is best from all the applied models. It also creates a weighted average for all the models.
Getting ready
You have completed all the recipes, and you have dataset X
,Y
, X_Hold
, and Y_Hold
created from the previous recipes.
How to do it...
Perform the following steps in R:
> install.packages("SuperLearner") > library(SuperLearner) > sl_models = c("SL.xgboost", "SL.randomForest", "SL.glmnet", "SL.nnet", "SL.rpartPrune", "SL.lm", "SL.mean") > superlearner = SuperLearner(Y = Y, X = X, family = gaussian (), SL.library = sl_models) > superlearner Output Call: SuperLearner(Y = Y, X = X, family = gaussian(), SL.library = sl_lib) Risk Coef SL.xgboost_All 7.606564e+00 0.000000e+00 SL.randomForest_All 1.027187e+01 2.907956e-16 SL.glmnet_All 8.641940e-02 0.000000e+00 SL.nnet_All 8.442002e+01 0.000000e...