Adjusting for non-linear relationship
The next three examples work with datasets where the underlying data structure violates the linear regression technical conditions. For each example, you will apply a transformation to the data in order to create residual plots that look scattered.
In this first example, it appears as though the variables are not linearly related.
This exercise is part of the course
Inference for Linear Regression in R
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Run this to see how the model looks
ggplot(hypdata_nonlinear, aes(x = explanatory, y = response)) +
geom_point() +
geom_smooth(method = "lm", se = FALSE)
# Model response vs. explanatory
model <- ___
# Extract observation-level information
modeled_observations <- ___
# See the result
modeled_observations
# Using modeled_observations, plot residuals vs. fitted values
___ +
# Add a point layer
___ +
# Add horizontal line at y = 0
___