Building a workflow
With your data ready for analysis, you will declare a logistic_model()
to predict whether or not they will arrive late.
You assign the role of "ID" to the flight
variable to keep it as a reference for analysis and debugging. From the date
variable, you will create new features to explicitly model the effect of holidays and represent factors
as dummy variables.
Bundling your model and recipe()
together using workflow()
will help ensure that subsequent fittings or predictions will implement consistent feature engineering steps.
This exercise is part of the course
Feature Engineering in R
Exercise instructions
- Assign an "ID" role to
flight
. - Bundle the model and the recipe into a
workflow
object. - Fit
lr_workflow
to thetest
data. - Tidy the fitted workflow.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
lr_model <- logistic_reg()
# Assign an "ID" role to flight
lr_recipe <- recipe(arrival ~., data = train) %>% update_role(flight, new_role = ___) %>%
step_holiday(date, holidays = timeDate::listHolidays("US")) %>% step_dummy(all_nominal_predictors())
# Bundle the model and the recipe into a workflow object
lr_workflow <- workflow() %>% add_model(___) %>% add_recipe(___)
lr_workflow
# Fit lr_workflow workflow to the test data
lr_fit <- lr_workflow %>% ___(data = test)
# Tidy the fitted workflow
tidy(___)