R: by group mirrored histogram using ggplot2

问题

I would like to get a mirrored histogram by group using ggplot as illustrated in the picture at the bottom.

library(ggplot2)
data2 <- data.frame(
  type = c( rep("Top 1", n_segment),
            rep("Top 2", n_segment),
            rep("Bottom 1", n_segment),
            rep("Bottom 2", n_segment)),
  value = c( rnorm(n_segment, mean=5),
             rnorm(n_segment, mean=12),
             rnorm(n_segment, mean=-5),
             rnorm(n_segment, mean=-12))
)

# Represent it
data2 %>%
  ggplot( aes(x=value, fill=type)) +
  geom_density( color="#e9ecef", alpha=0.6) 

For now I am getting this:

But I would like to mirror them as follows:

回答1:

Try this approach. You can work with ..density.. and adjust geom_density() around different subsets of data enabling data option and smartly filtering which values go up and down. Here the code:

library(ggplot2)
library(dplyr)
# Represent it
data2 %>%
  ggplot( aes(x=value, fill=type,group=type)) +
  geom_density(aes(y=-1*..density..),alpha=0.6,
               data = ~ subset(., type %in% c("Bottom 1","Bottom 2")))+
  geom_density(aes(y=..density..),alpha=0.6,
               data = ~ subset(., !type %in% c("Bottom 1","Bottom 2")))+
  ylab('density')

Output:

If you want to avoid lines, try this:

# Represent it 2
data2 %>%
  ggplot( aes(x=value, fill=type,group=type)) +
  geom_density(aes(y=-1*..density..),alpha=0.6,color='transparent',
               data = ~ subset(., type %in% c("Bottom 1","Bottom 2")))+
  geom_density(aes(y=..density..),alpha=0.6,color='transparent',
               data = ~ subset(., !type %in% c("Bottom 1","Bottom 2")))+
  ylab('density')

Output:

来源：https://stackoverflow.com/questions/64957319/r-by-group-mirrored-histogram-using-ggplot2