Visualize Scatterplots with Marginal Distribution using ggExtra

Marginal Distribution with ggplot2 and ggExtra

The data in this example is from the UN Statistics Division Sustainable Development Goal, Indicator 4.4.1.

Also check out the r-graph-gallery.com for inspiration.

Here’s the breakdown:

1. Load Packages and Libraries

The key here is the ggExtra package.

install.packages('ggExtra')
library(ggExtra)
library(tidyverse)

2. Create a basic scatter plot

The key here is using pivot_wider to give all Type of skill their own columns. We’ll then pick out specific columns (i.e., COPA, EMAIL, PCPR) to summarize, then plot on the x- and y- axes.

# Basic Scatter Plot (color cluster by Gender)
p <- data %>%
    select(GeoAreaName, TimePeriod, Sex, `Type of skill`, Value) %>%
    group_by(GeoAreaName, TimePeriod, Sex, `Type of skill`, Value) %>%
    pivot_wider(names_from = `Type of skill`, values_from = Value) %>%
    mutate(
        COPA = as.numeric(COPA),
        EMAIL = as.numeric(EMAIL),
        PCPR = as.numeric(PCPR)
    ) %>%
    # Group by GeoAreaName, across TimePeriod, Sex
    group_by(GeoAreaName, Sex) %>%
    summarize(
        avg_COPA = mean(COPA, na.rm = TRUE),
        avg_EMAIL = mean(EMAIL, na.rm = TRUE),
        avg_PCPR = mean(PCPR, na.rm = TRUE)
    ) %>%
    ungroup() %>%
    ggplot(aes(x = avg_PCPR, y = avg_EMAIL, color = Sex)) + 
    geom_point()

3. Use ggMarginal() to create the marginal distribution along the side of the scatter plots. This is a function from the ggExtra package.

# Scatter Plot with Marginal Distribution
ggMarginal(p, type = 'histogram')

This particular chart is especially useful to highlight different distributions.