Blend, fit, predict
blend_predictions() performs LASSO regularization to combine the outputs from the stack members to come up with one final prediction.- Candidates with non-zero coefficients are kept.
tree_frogs_model_st <-
tree_frogs_data_st %>%
blend_predictions()
- There’s an
autoplot() function available, to see what’s going on.
- If you don’t like what you’re seeing, you can try
blend_predictions() again, and setting your own penalty argument.
- Essentially, what you have, is a linear combination of each member’s prediction, to create one final prediction.
- With this “instruction” on how to combine candidate models, we fit the whole training set
tree_frogs_model_st <-
tree_frogs_model_st %>%
fit_members()
- And predict on testing set
tree_frogs_test <-
tree_frogs_test %>%
bind_cols(predict(tree_frogs_model_st, .))