Allowing for different colors for boxplot and overlying points in ggplot2-CodePudding

Here is the data:

set.seed(1234)
df = tribble(
    ~group,~value,
    "a",   rnorm(1000, mean = 5),
    "b",   rnorm(1000, mean = 7, sd = 1.5),
    "c",   rnorm(1000, mean = 8),
    "c",   rnorm(1000, mean = 9),
    "c",   rnorm(1000, mean = 7)
) %>%
    unnest(value)

I use the code below to create box plot and overlying dots:

library(ggplot2)
ggplot(df, aes(x = group, y = value))  
geom_boxplot(aes(color = group), fill = NA, outlier.shape = NA) 
geom_point(aes(color = group), shape = 21, position = position_jitter(
    seed = 1, width = .05
), alpha = 0.1)

Here is the figure:

The border color of the overlying points and the border color of the boxes are identical. For example, when the boxplot is orange, the border of points is also orange. I wish to know how could I allow for different colors for the boxplot and overlying points' border. For example, I wish to make colors of the boxplot to be black, blue, brown for groups a, b, and c, respectively. I also want the color of the border of overlying points to be red, green, yellow.

CodePudding user response：

Maybe you look for something like this:

Transform group to factor and then recode with recode_factor from dplyr package with colors you desire. Then within ggplot assign your new colors:

library(tidyverse)

df %>% 
  mutate(group = factor(group),
         color_group = recode_factor(group, "a"="gold", "b"="purple", "c"="black")) %>% 
  ggplot(aes(x = group, y = value))  
  geom_boxplot(aes(color = color_group), fill = NA, outlier.shape = NA) 
  geom_point(aes(color = group), shape = 21, position = position_jitter(
    seed = 1, width = .05
  ), alpha = 0.1)

CodePudding user response：

Another option to achieve your desired result would be to use the ggnewscale package which allows for multiple scales and legends for the same aesthetic:

library(ggplot2)
library(ggnewscale)

ggplot(df, aes(x = group, y = value))  
  geom_boxplot(aes(color = group), fill = NA, outlier.shape = NA)  
  scale_color_manual(values = c("black", "blue", "brown"))  
  new_scale_color()  
  geom_point(aes(color = group),
    shape = 21,
    position = position_jitter(seed = 1, width = .05), alpha = 0.1
  )  
  scale_color_manual(values = c("red", "green", "yellow"))