🗓️ Session 7: Visualization

One area for which R is particulary well-known for is the area of visualization. This is particularly because of the package ggplot2. This session introduces ggplot2 and the general approach to generate visualization in R. The good thing is that if you follow the approach described here, you can basically create every visualization type you can think of.

👨‍🏫 Lecture Slides

Either click on the slide area below or click here to download the slides.

The R script of this session

	library(DataScienceExercises)
	library(ggplot2)
	library(scales)
	# Bubble plot------
	gdp_data <- DataScienceExercises::gdplifexp2007
	# In the following, lines with changes are marked with '# <---'
	# 1st step: empty list
	gdp_plot <- ggplot()
	gdp_plot
	# 2nd step: add reference to underlying data set
	gdp_plot <- ggplot(data = gdp_data)
	gdp_plot
	# 3rd step: add aesthetic mappings
	gdp_plot <- ggplot(
	data = gdp_data,
	mapping = aes(x = gdpPercap, y = lifeExp) # <---
	)
	gdp_plot
	# 4th step: add point geometry
	gdp_plot <- ggplot(
	data = gdp_data,
	mapping = aes(x = gdpPercap, y = lifeExp)) +
	geom_point() # <---
	gdp_plot
	# 5th step: add size aesthetic
	gdp_plot <- ggplot(
	data = gdp_data,
	mapping = aes(
	y = lifeExp,
	color = continent,
	size = pop, # <---
	x = gdpPercap)
	) +
	geom_point()
	gdp_plot
	# 6th step: specify transparency of points and use different shape to
	# get difference between color and fill aesthetic
	# source for different shapes (scroll down to the bottom):
	# https://ggplot2.tidyverse.org/reference/aes_linetype_size_shape.html
	gdp_plot <- ggplot(
	data = gdp_data,
	mapping = aes(
	y = lifeExp,
	fill = continent, # <---
	size = pop,
	x = gdpPercap)
	) +
	geom_point(alpha=0.65, shape = 21) # <---
	gdp_plot
	# 7th step: modify the scales for fill, size, and x aesthetic (i.e. the way
	# the values are mapped on these aesthetics):
	gdp_plot <- ggplot(
	data = gdp_data,
	mapping = aes(
	y = lifeExp,
	fill = continent,
	size = pop,
	x = gdpPercap)
	) +
	geom_point(alpha=0.65, shape = 21) +
	scale_fill_brewer(palette = "Dark2") + # <---
	scale_size_continuous(range = c(0.1, 21)) + # <---
	scale_x_continuous(
	labels = label_number(scale = 0.001, suffix = "k") # <---
	)
	gdp_plot
	# 8th step: specify the labels for x and y axis, title, and add caption
	gdp_plot <- ggplot(
	data = gdp_data,
	mapping = aes(
	y = lifeExp,
	fill = continent,
	size = pop,
	x = gdpPercap)
	) +
	geom_point(alpha=0.65, shape = 21) +
	scale_fill_brewer(palette = "Dark2") +
	scale_size_continuous(range = c(0.1, 21)) +
	scale_x_continuous(
	labels = label_number(scale = 0.001, suffix = "k")
	) +
	labs( # <---
	x="GDP per capita", # <---
	y = "Life expectancy in years", # <---
	title = "Life expectancy and income per capita", # <---
	caption = "Data: Gapminder.") # <---
	gdp_plot
	# 9th step: Remove label for size aesthetic
	gdp_plot <- ggplot(
	data = gdp_data,
	mapping = aes(
	y = lifeExp,
	fill = continent,
	size = pop,
	x = gdpPercap)
	) +
	geom_point(alpha=0.65, shape = 21) +
	scale_fill_brewer(palette = "Dark2") +
	scale_size_continuous(range = c(0.1, 21), guide = "none") + # <---
	scale_x_continuous(
	labels = label_number(scale = 0.001, suffix = "k")
	) +
	labs(
	x="GDP per capita",
	y = "Life expectancy in years",
	title = "Life expectancy and income per capita",
	caption = "Data: Gapminder.")
	gdp_plot
	# 10th step: Use theme_bw() to fix background and other smaller plot issues
	# theme_bw() is one of the many gggplot2() themes that summarize many
	# changes to the function theme() into one call. See an overview, e.g., here:
	# https://ggplot2-book.org/themes#sec-themes
	gdp_plot <- ggplot(
	data = gdp_data,
	mapping = aes(
	y = lifeExp,
	fill = continent,
	size = pop,
	x = gdpPercap)
	) +
	geom_point(alpha=0.65, shape = 21) +
	scale_fill_brewer(palette = "Dark2") +
	scale_size_continuous(range = c(0.1, 21), guide = "none") +
	scale_x_continuous(
	labels = label_number(scale = 0.001, suffix = "k")
	) +
	labs(
	x="GDP per capita",
	y = "Life expectancy in years",
	title = "Life expectancy and income per capita",
	caption = "Data: Gapminder.") +
	theme_bw() + # <---
	theme(
	legend.position = "bottom", # <---
	legend.title = element_blank(), # <---
	panel.border = element_blank(), # <---
	axis.line = element_line(colour = "grey"), # <---
	axis.ticks = element_blank() # <---
	)
	gdp_plot
	# 11th step: Use theme() to fix position of the legend and remove legend title
	gdp_plot <- ggplot(
	data = gdp_data,
	mapping = aes(
	y = lifeExp,
	fill = continent,
	size = pop,
	x = gdpPercap)
	) +
	geom_point(alpha=0.65, shape = 21) +
	scale_fill_brewer(palette = "Dark2") +
	scale_size_continuous(range = c(0.1, 21), guide = "none") +
	scale_x_continuous(
	labels = label_number(scale = 0.001, suffix = "k")
	) +
	labs(
	x="GDP per capita",
	y = "Life expectancy in years",
	title = "Life expectancy and income per capita",
	caption = "Data: Gapminder.") +
	theme_bw() +
	theme(
	legend.position = "bottom", # <---
	legend.title = element_blank() # <---
	)
	gdp_plot
	# 12th step: Use theme() to fix the panel border and axis lines and ticks
	gdp_plot <- ggplot(
	data = gdp_data,
	mapping = aes(
	y = lifeExp,
	fill = continent,
	size = pop,
	x = gdpPercap)
	) +
	geom_point(alpha=0.65, shape = 21) +
	scale_fill_brewer(palette = "Dark2") +
	scale_size_continuous(range = c(0.1, 21), guide = "none") +
	scale_x_continuous(
	labels = label_number(scale = 0.001, suffix = "k")
	) +
	labs(
	x="GDP per capita",
	y = "Life expectancy in years",
	title = "Life expectancy and income per capita",
	caption = "Data: Gapminder.") +
	theme_bw() +
	theme(
	legend.position = "bottom",
	legend.title = element_blank(),
	panel.border = element_blank(), # <---
	axis.line = element_line(colour = "grey"), # <---
	axis.ticks = element_blank() # <---
	)
	gdp_plot
	# Line plot------
	gdp_data_time <- DataScienceExercises::aggGDPlifexp
	# The following code is copy-pasted from above, with changes being marked by comments
	gdp_line_plot <- ggplot(
	data = gdp_data_time, # Change data set
	mapping = aes(# Adjust mappings
	y = gdpPercap,
	color = continent,
	x = year)
	) +
	geom_point(alpha=0.65) + # Remove shape specification
	geom_line() + # Add lines to the plot
	scale_color_brewer(palette = "Dark2") + # Change fill to color
	# scale_size_continuous( # Not necessary any more
	# range = c(0.1, 21), guide = "none") +
	scale_y_continuous(# Change from scale_x
	labels = scales::label_number(scale = 0.001, suffix = "k")
	) +
	labs(# Adjust labels
	y="GDP per capita",
	title = "Divergences in income",
	caption = "Data: Gapminder.") +
	theme_bw() +
	theme(
	legend.position = "bottom",
	legend.title = element_blank(),
	panel.border = element_blank(),
	axis.line = element_line(colour = "grey"),
	axis.ticks = element_blank(),
	axis.title.x = element_blank() # Remove title for x axis
	)
	gdp_line_plot

view raw VisualizationScript.R hosted with ❤ by GitHub

🎥 Lecture videos

Currently, no videos are available for this session. Check out the tutorial instead.

📚 Mandatory Reading

• The tutorial visualization

Further Reading

• Wickham (2010), who introduces the theory underlying ggplot2
• ggplot2 cheat sheet
• Browse the website from Data to Viz and try to re-create some of the figures yourself

✍️ Coursework

• Do the exercises Visualization1 from the DataScienceExercises package

Quick code for starting the exercises

learnr::run_tutorial(
  name = "Visualization1", 
  package = "DataScienceExercises", 
  shiny_args=list("launch.browser"=TRUE))

References

Wickham, H. (2010) “A Layered Grammar of Graphics,” Journal of Computational and Graphical Statistics 19(1): 3–28.