A stacked range plot using ggplot - r

Question

Suppose I wish to make a range plot with the design below using ggplot with the following dummy data:
with following legend.
set.seed(1)
test.dat <- data.frame(
yval = sample(1:100, 40),
xcat = rep(LETTERS[1:4], 10),
base = sample(c(1, 0),40, replace=T),
col = rep(c("red", "blue"), 40)
)
> head(test.dat)
yval xcat base col
1 68 A 0 red
2 39 B 0 blue
3 1 C 0 red
4 34 D 1 blue
5 87 A 0 red
6 43 B 0 blue
The gray portion shows the range of the data where base == 1 and the whisker-like line (that resembles errorbar) shows the range of the data where base == 0 using the respective color designed for each xcat.
So using this dummy data, I would expect:
minmax <- function(x){
return(
c(min(x),max(x))
)
}
> minmax(test.dat[test.dat$xcat == "D" & test.dat$base == 1,]$yval)
[1] 24 99
> minmax(test.dat[test.dat$xcat == "D" & test.dat$base == 0,]$yval)
[1] 21 82
> unique(test.dat[test.dat$xcat == "D",]$col)
[1] "blue"
for xcat == "D", a gray bar to range from 24 to 99, and a blue whisker line to range from 21 to 82.
How can I achieve this? It looks like there is no straightforward ggplot function to create a range plot.
My approach idea was to adjust geom_boxplot's hinges and whisper definition for gray part, and use geom_line or geom_linerange to create the whisker-line part, but I am unsure how to do that.
Thank you.

Answer 1

You first create a dataframe where you have min and max for each combination of (xcat, base and col)
data2 <- test.dat %>% group_by(xcat, base, col) %>% summarise(min = min(yval), max=max(yval))
Then you use geom_linerange for the gray "bars" and geom_errorbar for the whisker line:
ggplot()+
geom_linerange(data= data2 %>% filter(base==1), aes(x= xcat, ymin=min, ymax=max), size=12, alpha=0.5)+
geom_errorbar(data= data2 %>% filter(base==0), aes(x= xcat, ymin=min, ymax=max), colour=data2[data2$base==1,]$col, width=.2)
And this is the
Plot

Answer 2

I would suggest doing some reshaping first using dplyr/tidyr, and then geom_tile:
library(tidyverse)
test.dat %>%
group_by(xcat, base, col) %>%
summarize(mid = mean(range(yval)),
range = diff(range(yval)), .groups = "drop") %>%
pivot_wider(names_from = base, values_from = mid:range) %>%
ggplot(aes(x = xcat)) +
geom_tile(aes(y = mid_0, height = range_0), fill = "gray70", color = "black") +
geom_tile(aes(y = mid_1, height = range_1, fill = col), color = "black") +
scale_fill_identity()