ggplot2: grouped bar graph with groupby results as one bar and the total as second bar-CodePudding

I have a dataframe as follows

df = data.frame(col1 = c('a', 'b', 'c','d','e'), col2 = c(3,1,2,3,1), col3 = c('yes','yes','no','no','yes))

Essentially I would like a grouped bar chart where the first group is the number of times a value occurs in column2 and has a value of yes in column3, and the second bar is the total number of times the value occurs in column2, regardless of group.

CodePudding user response：

Use a grouped summary to compute yes and total, pivot to long, and map the pivoted column to fill:

library(tidyverse)

df %>%
  group_by(col2) %>%
  summarize(
    yes = sum(col3 == "yes"),
    total = n()
  ) %>%
  pivot_longer(
    yes:total,
    names_to = "response",
    values_to = "n"
  ) %>%
  mutate(response = fct_rev(response)) %>%
  ggplot()  
  geom_col(
    aes(col2, n, fill = response),
    position = "dodge"
  )  
  scale_y_continuous(limits = c(0, 3), expand = c(0, 0))  
  scale_fill_manual(values = c("#7570b3", "#1b9e77"))  
  theme_classic()