Showing Top & Bottom values in one clear visualization

A simple pattern to compare extremes of a distribution

data-wrangling
functions
R-Hacks N.4
Author

Lucio Colonna

Published

January 23, 2026

This hack is based on my analysis on Kaggle analysis linked as follows (please see Chapter 4.3):
🔗 https://www.kaggle.com/code/lcolon/exploring-2024-software-engineer-salaries


When exploring a distribution, a very common question is:

Who are the top performers, and who are at the bottom?

Showing only the Top values might hid some context, while showing two separate charts can make comparison harder.

This R-Hack shows a simple and reusable pattern to display Top and Bottom values together in a single, clean visualization, as the one showed below:




Step 0 – Create an example dataset

Let’s start from a small, simulated dataset representing average salaries for different companies:

library(tidyverse)

set.seed(123)

df <- data.frame(
  company = (LETTERS[1:26]),
  avg_salary = round(runif(26, min = 60, max = 180), 0)
)

head(df)
  company avg_salary
1       A         95
2       B        155
3       C        109
4       D        166
5       E        173
6       F         65


Step 1 – Build Top & Bottom datasets in a single pipeline

In this step, we extract both the Top 10 and Bottom 10 values using a single, readable pipeline.

The idea is simple. Starting from the same dataset:

  • select the Top 10 by applying slice_max(), which returns all observations within the top 10 ranking positions, including any ties

  • select the Bottom 10 by applying slice_min(), returning all observations within the lowest 10 ranking positions, again including ties

  • assign a clear group label to each subset

  • combine the two subsets into a single dataframe for visualization and further analysis

plot_df <- 
  bind_rows(
    df %>% 
      slice_max(avg_salary, n = 10, with_ties = TRUE) %>%
      mutate(group = "Top 10"),
    
    df %>% 
      slice_min(avg_salary, n = 10, with_ties = TRUE) %>%
      mutate(group = "Bottom 10")
  )

head(plot_df)
  company avg_salary  group
1       X        179 Top 10
2       K        175 Top 10
3       T        175 Top 10
4       E        173 Top 10
5       P        168 Top 10
6       H        167 Top 10


Step 2 - Plot the data

In this step, starting from the dataframe created in the previous step, we build a simple visualization with ggplot:

plot_base <- ggplot(plot_df, 
                    aes(x = reorder(company, -avg_salary), y = avg_salary)) +
  geom_col(fill = "steelblue") +
  geom_text(aes(label = avg_salary), vjust = -0.4, size = 4) +
  facet_wrap(~ fct_rev(group), scales = "free_x") +
  labs(
    title = "Top and Bottom Companies by Average Salary",
    subtitle = "Simulated data example\n\n",
    y = "Average yearly salary",
    x = NULL
  ) +
  theme_minimal() +
  theme(
    panel.grid = element_blank(),
    axis.text.x = element_text(size = 10),
    axis.text.y = element_blank()
  )

plot_base


Step 3 - Generate gradient color palettes

Now that the base chart is working, we can make it more informative by adding two gradient color palettes:

  • one gradient for the Top group (blue shades)
  • one gradient for the Bottom group (red shades)

Instead of assigning a single color per group, we assign a slightly different shade to each bar. This creates a clean gradient effect while keeping the plot readable — even when ties produce more than 10 observations per group.

To do this, we:

  • generate a group-specific palette with colorRampPalette(), sized to the number of rows in each group (so it adapts if ties expand the selection)
  • group the data by group (Top vs Bottom)
  • assign colors row by row using row_number()
  • store the result in a new column called color 
plot_df <- plot_df %>%
  group_by(group) %>%
  mutate(
    color = if_else(
      group == "Top 10",
      colorRampPalette(c("darkblue", "lightblue"))(n())[row_number()],
      colorRampPalette(c("darkred", "lightcoral"))(n())[row_number()]
    )
  ) %>%
  ungroup()

head(plot_df)
# A tibble: 6 × 4
  company avg_salary group  color  
  <chr>        <dbl> <chr>  <chr>  
1 X              179 Top 10 #00008B
2 K              175 Top 10 #131895
3 T              175 Top 10 #26309F
4 E              173 Top 10 #3948A9
5 P              168 Top 10 #4C60B3
6 H              167 Top 10 #6078BD


Step 4 - Style the chart using the new colors

At this point, we don’t want to rewrite the entire ggplot call. Instead, we:

  • reuse the base chart (plot_base)

  • replace its dataset with the updated plot_df (the one that now includes color) using the %+% operator

  • add a new geom_col() that maps fill to the color column

  • use scale_fill_identity() so ggplot uses the colors as they are listed in the color column

plot_styled <- (plot_base %+% plot_df) +
  geom_col(aes(fill = color)) +
  scale_fill_identity()

plot_styled


In short

  • Extract Top 10 and Bottom 10 values from the same dataset using slice_max() and slice_min(), including ties

  • Combine the two subsets into a single dataframe for a compact overview of the distribution extremes

  • Build a clean base plot to validate structure and layout

  • Enhance the visualization by applying gradient colors to distinguish Top and Bottom groups


Tip

If you want to stay up to date with the latest events from the Rome R Users Group, click here:

👉 https://www.meetup.com/rome-r-users-group/

And if you are curious, the full Kaggle notebook used for this tip is available here:

🔗 https://www.kaggle.com/code/lcolon/exploring-2024-software-engineer-salaries

Back to top