Fine tune your model
Wow! That was a significant improvement over a regression model. Now let's see if you can further improve this performance by fine tuning your random forest models. To do this you will vary the mtry
parameter when building your random forest models on your train
data.
The default value of mtry
for ranger is the rounded down square root of the total number of features (6). This results in a value of 2.
This exercise is part of the course
Machine Learning in the Tidyverse
Exercise instructions
- Use
crossing()
to expand the cross validation data for values ofmtry
ranging from 2 through 5. - Build random forest models for each fold/mtry combination.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Prepare for tuning your cross validation folds by varying mtry
cv_tune <- cv_data %>%
crossing(mtry = ___:___)
# Build a model for each fold & mtry combination
cv_model_tunerf <- cv_tune %>%
mutate(model = map2(.x = ___, .y = ___, ~ranger(formula = life_expectancy~.,
data = .x, mtry = .y,
num.trees = 100, seed = 42)))