Ordering stacked bar charts using ggplot

Photo by Eaters Collective on Unsplash

In data visualization, stacked bar charts are a useful way to represent categorical data with multiple subcategories. However, the default ordering of the bars might not always be optimal for conveying the desired message. In this post, we will explore how to order stacked bar charts using the popular ggplot2 package in R.

Let’s fabricate a fake dataset first:

data <- data.frame(
    country = c("Country A", "Country B", "Country C", "Country D", "Country E"),
    Agriculture = c(82.2, 78.4, 73.2, 60.3, 57.8),
    Education = c(1.63, 3.29, 3.65, 6.56, 8.94),
    Healthcare = c(11.0, 10.7, 15.2, 15.1, 16.0),
    IT = c(5.18, 7.61, 7.91, 18.00, 17.30)
)

> data
    country Agriculture Education Healthcare    IT
1 Country A        82.2      1.63       11.0  5.18
2 Country B        78.4      3.29       10.7  7.61
3 Country C        73.2      3.65       15.2  7.91
4 Country D        60.3      6.56       15.1 18.00
5 Country E        57.8      8.94       16.0 17.30

Our goal is to visualize and compare the percent distribution of employment by sector.

To do so, the data will take a quick reshaping:

data = data %>%
    mutate(country = factor(country, 
                           levels = country[order(IT, decreasing = F)])) %>%
    pivot_longer(cols = -country, names_to = "Frequency", values_to = "Percentage") 


# A tibble: 20 × 3
   country   Frequency   Percentage
   <fct>     <chr>            <dbl>
 1 Country A Agriculture      82.2 
 2 Country A Education         1.63
 3 Country A Healthcare       11   
 4 Country A IT                5.18
 5 Country B Agriculture      78.4 
 6 Country B Education         3.29
 7 Country B Healthcare       10.7 
 8 Country B IT                7.61
 9 Country C Agriculture      73.2 
10 Country C Education         3.65
11 Country C Healthcare       15.2 
12 Country C IT                7.91
13 Country D Agriculture      60.3 
14 Country D Education         6.56
15 Country D Healthcare       15.1 
16 Country D IT               18   
17 Country E Agriculture      57.8 
18 Country E Education         8.94
19 Country E Healthcare       16   
20 Country E IT               17.3

to generate the desired plot:

data %>%
    mutate(country = factor(country, 
                            levels = country[order(IT, decreasing = F)])) %>%
    pivot_longer(cols = -country, names_to = "Frequency", values_to = "Percentage") %>%
    ggplot(aes(x = country, y = Percentage, fill = Frequency)) +
    geom_bar(stat = "identity") + theme_minimal()+
    coord_flip()