Yeo-Johnson transformation
Using the attrition_num
dataset with all numerical data about employees who have left the company, you want to build a model that can predict if an employee is likely to stay, using Attrition
, a binary variable coded as a factor
. To get the features to behave nearly normally, you will create a recipe that implements the Yeo-Johnson transformation.
The attrition_num
data, the logistic regression lr_model
, the user-defined class-evaluate()
function, and the train
and test
splits are loaded for you.
This exercise is part of the course
Feature Engineering in R
Exercise instructions
- Create a recipe that uses Yeo-Johnson to transform all numeric features, including the target.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Create a recipe that uses Yeo-Johnson to transform all numeric features
lr_recipe_YJ <-
recipe(Attrition ~., data = train) %>%
___
lr_workflow_YJ <- workflow() %>%
add_model(lr_model) %>%
add_recipe(lr_recipe_YJ)
lr_fit_YJ <- lr_workflow_YJ %>%
fit(train)
lr_aug_YJ <-
lr_fit_YJ %>% augment(test)
lr_aug_YJ %>% class_evaluate(truth = Attrition,
estimate = .pred_class,.pred_No)