Starwars (2)

library(tidyverse)

starwars_filtered <- starwars %>%
  filter(!is.na(mass), !is.na(height))

Create a scatterplot of mass and height in the starwars dataset

ggplot(data = starwars_filtered, aes(x = height, y = mass)) +
  geom_point() + 
  labs(title = "Relationship between height and mass of Starwars characters")

Fit a model of mass ~ height using the starwars dataset

model <- lm(mass ~ height, starwars_filtered)
model

## 
## Call:
## lm(formula = mass ~ height, data = starwars_filtered)
## 
## Coefficients:
## (Intercept)       height  
##    -13.8103       0.6386

Calculate the studentized residuals and add them to the starwars dataset

starwars_filtered <- starwars_filtered %>%
  mutate(student_residual = rstudent(model))

Create a plot of the studentized residuals on the y-axis and height on the x-axis

ggplot(starwars_filtered, aes(height, student_residual)) + 
  geom_point() +
  geom_hline(yintercept = c(-4, -2, 2, 4), lty = 2) + 
  labs(title = "Studentized residuals")

Fit the same regression model (mass ~ height) with and without the “unusual” point

starwars %>%
  filter(name != "Jabba Desilijic Tiure") %>%
  lm(mass ~ height, data = .)

## 
## Call:
## lm(formula = mass ~ height, data = .)
## 
## Coefficients:
## (Intercept)       height  
##    -32.5408       0.6214