Build a random forest model

Here you will use the same cross-validation data to build (using train) and evaluate (using validate) random forests for each partition. Since you are using the same cross-validation partitions as your regression models, you are able to directly compare the performance of the two models.

Note: We will limit our random forests to contain 100 trees to ensure they finish fitting in a reasonable time. The default number of trees for ranger() is 500.

This exercise is part of the course

Machine Learning in the Tidyverse

View Course

Exercise instructions

Use ranger() to build a random forest predicting life_expectancy using all features in train for each cross validation partition.
Add a new column validate_predicted predicting the life_expectancy for the observations in validate using the random forest models you just created.

Hands-on interactive exercise

Have a go at this exercise by completing this sample code.

library(ranger)

# Build a random forest model for each fold
cv_models_rf <- cv_data %>% 
  mutate(model = map(___, ~ranger(formula = ___, data = ___,
                                    num.trees = 100, seed = 42)))

# Generate predictions using the random forest model
cv_prep_rf <- cv_models_rf %>% 
  mutate(validate_predicted = map2(.x = ___, .y = ___, ~predict(.x, .y)$predictions))

Edit and Run Code