How to change the colour for specific points in a Cook's distance plot?

Question

I have plotted Cook's distance via the car package:

 library(car)
 influenceIndexPlot(model, id.n = 5, vars = c("Cook"))

http://gyazo.com/4b15671a0ee35944c4b5b273205f1b72

I would like to draw all points and the lines of these points in red, which are above 0.01.

Is there a way how I could do that? Thank you very much in advance!

Answer 1

I'm not sure how to use that particular function from the car package to do what you are asking. It is just a wrapper to make a base-r plot.

here, I'm showing you how to make the same sort of plot in ggplot2.

Example: make some sample data and run a linear model:

set.seed(84)
df <- data.frame(x = rnorm(100, 10, 5), y = rnorm(100, 12, 5))

model <- lm(y ~ x, df)

now we get the Cook's distances, create a dataframe and assign groups - either 0 (below 0.01) or 1 (above 0.01):

N <- nrow(model$model) #this is just to get the number of observations - it's obvious it's 100 as we chose that, but put it here for automation

df <- data.frame(Index = 1:N,  Cook = cooks.distance(model))
df$group <- factor(ifelse(df$Cook > 0.01, 1, 0))

Now we plot it:

library(ggplot2)

ggplot(df, aes(Index, Cook, color=group, group=group)) +
  geom_point(size=3) +
  geom_segment(aes(Index, xend=Index, 0, yend=Cook, color=group), data=df)  +
  theme_bw() +
  scale_color_manual(values=c("black", "red1")) +
  ylab("Cook's Distance") +
  ggtitle("Diagnostic Plots") +
  theme(legend.position = "none")

It looks like this: