ggplot geom_point() with colors based on specific, discrete values

Question

I am trying to plot data points with three different colors for three value ranges. For example:

library(ggplot2)
ggplot(mtcars, aes(wt, mpg)) + geom_point(aes(colour = qsec))

The above produces:

Now, I would like to modify this so that qseq values <17 are black, values between 17 and 19 are yellow and values above 19 are red. I've tried various approaches, but none of them seems to work:

• Taken from here

    ggplot(mtcars, aes(wt, mpg)) + geom_point(aes(fill = qsec)) + 
    scale_fill_gradient(colours = c("black","yellow","red"), 
    breaks=c(0,17,19), labels = format(c("0","17","19")))
  

This produces:

So, the colorbar seems correct but the colors are not actually applied.

• The solutions given in ggplot2 heatmap with colors for ranged values seem promising but I can't get them to work with geom_point().

• Same goes for various similar posts such as those listed in the sidebar.

I realize I will probably need to use some kind of discrete scale instead of scale_fill_gradientn but my attempts to use scale_color_manual() fail:

ggplot(mtcars, aes(wt, mpg)) + geom_point(aes(color = factor(qsec))) + 
scale_color_manual(values=c("black", "yellow","red")
Error: Insufficient values in manual scale. 30 needed but only 4 provided.

I am guessing I will somehow have to use cut() or factor() but I can't seem to figure out how. Any suggestions?

Solution

You need to cut your values into intervals:

library(ggplot2)
ggplot(mtcars, aes(wt, mpg)) + 
  geom_point(aes(colour = cut(qsec, c(-Inf, 17, 19, Inf))),
             size = 5) +
  scale_color_manual(name = "qsec",
                     values = c("(-Inf,17]" = "black",
                                  "(17,19]" = "yellow",
                                  "(19, Inf]" = "red"),
                     labels = c("<= 17", "17 < qsec <= 19", "> 19"))