How to add geom_segment to geom_density_ridges_gradient?

Question

I would like to add vertical segments to a ridgeline plot whose histograms show customized quantiles.

I managed to get the vertical segments if I map fill color with ..x... But I would like to show quantiles in the density plots. I wrote the following code:

library(datasets)
library(ggplot2)
data("iris")

iris_lines <- data.frame(Species = c("setosa", "versicolor", "virginica"),
                         x0 = c(5, 5.9, 6.5))

Figure1 <- ggplot(iris, aes(x=Sepal.Length, y=Species, fill=(..quantile..))) +
  geom_density_ridges_gradient(jittered_points = FALSE, calc_ecdf = TRUE, quantile_lines = c(TRUE), quantiles =c(0.1,0.25,0.75,0.9),scale=0.9, color='white')+
  geom_segment(data = iris_lines, aes(x = x0, xend = x0, y = as.numeric(Species), yend = as.numeric(Species) + c(.9,.5,.5)), color = "red") + scale_y_discrete(expand = c(0.01, 0))
Figure1

The code works if I map fill color as fill = ..x.. I get three vertical lines representing the mean of each density plot; however, if I map fill color as fill = ..quantile.. I get the following error:

Error in data.frame(..., check.names = FALSE) : 
  arguments imply differing number of rows: 1, 3

Welcome to SO. From your comment to the answer below I take it that the answer was helpful. Please, take the time to mark the answer as helpful by up-voting it, and/or by marking it as the "right" answer. — KoenV, Jun 18 '19 at 07:47

Jon Spring · Accepted Answer · 2019-06-18T02:57:41.160

Nice chart!

Add inherit.aes = F to the second geom so it doesn't try to match your data with the fill calculation in the ggplot(aes() call.

Figure1 <- ggplot(iris, aes(x=Sepal.Length, y=Species, fill=(..quantile..))) +
      geom_density_ridges_gradient(jittered_points = FALSE, 
                                   calc_ecdf = TRUE, 
                                   quantile_lines = c(TRUE), 
                                   quantiles =c(0.1,0.25,0.75,0.9),
                                   scale=0.9, color='white') +
      geom_segment(data = iris_lines, 
                   aes(x = x0, xend = x0, 
                       y = as.numeric(Species), yend = as.numeric(Species) + c(.9,.5,.5)),
                   color = "red", inherit.aes = F) +   #### HERE ####
      scale_y_discrete(expand = c(0.01, 0))
Figure1

Edit:

OP asked in comment about selectively labeling some elements and adding a label for the median line. Here's an approach, probably not the pithiest.

Figure1 <- ggplot(iris, aes(x=Sepal.Length, y=Species, 
                            fill = (..quantile..), 
                            color = (..quantile..))) +
  geom_density_ridges_gradient(jittered_points = FALSE, 
                               calc_ecdf = TRUE, 
                               quantile_lines = c(TRUE), 
                               quantiles =c(0.1,0.25,0.75,0.9),
                               scale=0.9, color='white') +
  geom_segment(data = iris_lines, 
               aes(x = x0, xend = x0, fill = "median",
                   y = as.numeric(Species), 
                   yend = as.numeric(Species) + c(.9,.5,.5),
                   color = "median")) +   #### HERE ####
  scale_y_discrete(expand = c(0.01, 0)) +

  scale_color_manual(name = "quantile",
                     limits = c(1:3, "median"),
                     values = alpha("firebrick1", c(0, 0, 0, 1)),
                     labels = c("<10%", "10-25%", "IQR", "median")) +
  scale_fill_manual(name = "quantile",
    limits = c(1:3, "median"),
    values = c("cadetblue", "coral", "orange", "white"), 
    na.value = "gray30",
    labels = c("<10%", "10-25%", "IQR", "median"))
Figure1

Jon, I was stuck for a few days, thanks a lot for helping me out with this!, I didn't know about `inherit.aes`. Just a quick question regarding the label. Is it possible to show only the first three elements of the label and add a fourth element for the vertical red lines? — Andres, Jun 18 '19 at 01:42
It's possible, but I don't know a simple way. See edited answer for a way using manual scales. Note the fill and color scales need to share a name to be shown on a combined basis. — Jon Spring, Jun 18 '19 at 02:59

How to add geom_segment to geom_density_ridges_gradient?

1 Answers1