How can I extend the quantile regression lines geom_quantile to forecast in ggplot?

Question

I am trying to plot the quantile regression lines for a set of data. I would like to extend the quantile regression lines from geom_quantile() in order to show how they forecast similar to using stat_smooth() with the fullrange argument set to TRUE. However, there is no fullrange argument for geom_quantile(). As an example, see below:

data("mpg")
library(ggplot2)
library(dplyr)

m <-
   ggplot(mpg, aes(displ,1/ hwy)) +
   geom_point()
m + geom_quantile() +
   scale_x_continuous(limits = c(1,9),
                      breaks = waiver(),
                      n.breaks = 8)

p <-
   ggplot(mpg, aes(displ,1/ hwy)) +
   geom_point()
p + stat_smooth(method = lm, fullrange = TRUE, se = FALSE, color = "red") +
   scale_x_continuous(limits = c(1,9),
                      breaks = waiver(),
                      n.breaks = 8)

m1 <-
   ggplot(mpg, aes(displ,1/ hwy)) +
   geom_point()
m1 + geom_quantile(fullrange = TRUE) +
   scale_x_continuous(limits = c(1,9),
                      breaks = waiver(),
                      n.breaks = 8)

The first part for m gives the quantile regression lines over the set of data. For p, I can show the linear regression line forecast out to a displacement of say 9. For m1 though, I can't extend out the regression lines. Would there be a way for me to tell ggplot to do this kind of forecast? Easier is preferable, of course, but I'll consider any suggestions. Thanks in advance!

Relevant github issue: https://github.com/tidyverse/ggplot2/issues/3190 — markus, Feb 17 '22 at 20:10

score 2 · Accepted Answer · answered Feb 17 '22 at 20:24

Under the hood, geom_quantile uses quantreg::rq, and it's very straightforward to use it directly to produce the same effect using geom_abline:

mod  <- quantreg::rq(I(1/hwy) ~ displ, tau = c(0.25, 0.5, 0.75), data = mpg)
r_df <- setNames(as.data.frame(t(coef(mod))), c("intercept", "gradient"))

m1 + geom_abline(data = r_df, aes(slope = gradient, intercept = intercept)) +
     scale_x_continuous(limits = c(1,9),
                      breaks = waiver(),
                      n.breaks = 8)

How can I extend the quantile regression lines geom_quantile to forecast in ggplot?

1 Answers1