Who's staying?
It's time to practice combining several transformations to the attrition_num
data. First, normalize or near-normalize numeric variables by applying a Yeo-Johnson transformation. Next, transform numeric predictors to percentiles, create dummy variables, and eliminate features with near zero variance.
This exercise is part of the course
Feature Engineering in R
Exercise instructions
- Apply a Yeo-Johnson transformation to all numeric variables.
- Transform all numeric predictors into percentiles.
- Create dummy variables for all nominal predictors.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
lr_recipe <- recipe(Attrition ~., data = train) %>%
# Apply a Yeo-Johnson transformation to all numeric variables
___ %>%
# Transform all numeric predictors into percentiles
___ %>%
# Create dummy variables for all nominal predictors
___
lr_workflow <- workflow() %>% add_model(lr_model) %>% add_recipe(lr_recipe)
lr_fit <- lr_workflow %>% fit(train)
lr_aug <- lr_fit %>% augment(test)
lr_aug %>% class_evaluate(truth = Attrition, estimate = .pred_class,.pred_No)