Barchart with 2 bars each group

December 11, 2023

I want to plot mean_1 and mean_2 separately by each value of grouping variable x.
Thus, I need two bars for each group.

Data:

n <- 100
set.seed(123)  
data <- data.frame(
  x = round(runif(n, min = 3, max = 5)),
  mean_1 = rnorm(n, mean = 0, sd = 1),
  mean_2 = rnorm(n, mean = 0, sd = 1) + 0.3)

Approach which just shows 1 bar:

library(ggplot2)

ggplot(data, aes(x = factor(x))) +
  geom_bar(aes(y = mean_1), stat = "identity") +
  geom_bar(aes(y = mean_2), stat = "identity") +
  labs(x = "x",
       y = "Abs") +
  scale_fill_manual(values = c("blue", "red")) +
  theme_minimal()

Thank you

>Solution :

You should first make your dataset tidy by using tidyr::pivot_longer (or any of your favorite reshaping approaches) then use ggplot, with a few tweaks to your existing code (see below)
Data

data_long <- tidyr::pivot_longer(data, mean_1:mean_2, 
                                 names_to = "mean", 
                                 values_to = "value")
#       x mean     value
#   <dbl> <chr>    <dbl>
# 1     4 mean_1  0.253 
# 2     4 mean_2  1.09  
# 3     5 mean_1 -0.0285
# 4     5 mean_2  1.07  
# ....

To plot, use y = value and fill = mean in the aes(), as well as position = position_dodge() in geom_bar:

ggplot(data_long, aes(x = factor(x), y = value, fill = mean)) +
  geom_bar(stat = "identity",  position = position_dodge()) +
  labs(x = "x",
       y = "Abs") +
  scale_fill_manual(values = c("blue", "red")) +
  theme_minimal()

The above plot shows the range of mean_1 and mean_2, so if that’s the goal of visualization you are all set. Though, with these data, you may be better off using geom_boxplot, which shows several summary statistics by group via the box, whiskers, and points:

ggplot(data_long, aes(x = factor(x), y = value, fill = mean)) +
  geom_boxplot() +
  labs(x = "x",
       y = "Abs") +
  scale_fill_manual(values = c("blue", "red")) +
  theme_minimal()

As @AllanCameron mentions, you may want to look into ... + geom_bar( stat = "summary", fun = mean, ...) + ..., which calculates and graphs the mean of each mean_1 and mean_2:

ggplot(data_long, aes(x = factor(x), y = value, fill = mean)) +
  geom_bar(stat = "summary", fun = mean, position = position_dodge()) +
  labs(x = "x", y = "Abs") +
  scale_fill_manual(values = c("blue", "red")) +
  theme_minimal()

Which produces the following plot: