A Grammar of Graphics

STAT 20: Introduction to Probability and Statistics

Concept Questions

1. What are the aesthetics and geometry of this plot?

01:00

ggplot(penguins, aes(x = bill_length_mm, y = species)) + geom_boxplot()

Correct answer: Bill length is mapped to the x, species is mapped to the y, and it uses the boxplot geometry.

This is a warm up to get them thinking about the fundamental concept of aesthetic mappings and geometries on a plot style and data set they’ll be familiar with. This should be quite easy.

What code was used to make this plot?

01:00

Correct answer: ggplot(penguins, aes(x = island, fill = island)) + geom_bar()

This demonstrates a practice called “double-mapping”, when the same column is mapped to two different aesthetic attributes. Most consider this a bad practice.

In a setting like this, usually the author is really just trying to get more color onto the plot. That can be accomplished by setting the color of the bars using something like:

Global infectious disease prevalence in 1989

2. What are the aesthetics and geometry of this plot?

01:00

Correct answer: Color is mapped to taxonomic family, size is mapped to prevalence, the diseases are represented using the point geometry.

Good follow up questions and boardwork include:

• What does the data frame look like that made this plot? (taxonomy is a categorical variable with three categories: bacteria, virus, and protoza, prevalence is proportion of population affected by disease, seems that disease is the observation so each point is a disease.)

• What determines where the bubbles are placed in the chart? Answer: bubble charts have different methods of doing this. This one appears to use a packing algorithm so get them close together. Important, none of those methods refer to a column in the data frame therefore are not aesthetic mappings.

• nice example of bubble chart that might appeal at https://www.mekkographics.com/top-english-premier-league-goal-scorers/

• If you really want to go far afield, you could show an example of a beeswarm chart, which is a bubble chart that uses one of the x or the y to map a numerical variable and then uses a packing algorithm to determine placement on the other axis. Here is one showing [Sherlock Holmes canon and adaptations]https://public.tableau.com/app/profile/jessica.bautista/viz/SherlockHolmesCanon/Beeswarm

• One of the lessons of this bubble chart is that our “named” charts correspond to particular combos of aesthetic mappings and geometriess. A “scatterplot” uses the point geometry and maps numerical data to the x and y axes. A “bubble chart” uses the point geometry too, but maps size to a numerical variable (and often uses color too), so adds two more variables.

Concept Activity

Concept Activity

You will be watching a 2.5 minute video of a presentation by a scientist, Hans Rosling, who studied global public health. He presents data visualizations depicting the change in life expectancy and family size over several decades in the 20th century.

On a piece of note paper:

• Sketch out the data frame used to create the graphic and add the names of the variables.
• List the aesthetic attributes used to encode the data in the graphic.
• Identify the geometry used in the plot.

If you don’t have much time, begin the video at 2:19 and stop at 4:53. If you have a bit of time, begin the video at the beginning, but still, stop at 4:53.

Please turn to your neighbors and…

Discuss what you came up with in terms of . . .

• the variables present in the data frame
• the aesthetic attributes used to encode that data in the plot
• the geometry

01:00

Change timer to at least 5:00. Then go on to the next slide, but pause with the slide still blank. Ask them what were the variables and aesthetic attributes, and write them down on the board.

What were the variables and aesthetic attributes?

Visual Cues / Aesthetics

• Location along the x-axis
• Location along the y-axis
• Size of point
• Color of point
• Animation

Variables

• Fertility rate
• Life expectancy
• Population
• Region
• Year

What did the data frame look like?

What was the unit of observation? What were the variables? What were their type?

Unit of observation

• A country in a given year

Variables

• Fertility rate (continuous)
• Life expectancy (continuous)
• Population (continuous)
• Region (nominal)
• Year (discrete)

What geometry is used to represent the observations?

• Points

Break

05:00

Worksheet: A Grammar of Graphics

20:00

Lab 2.2: Class Survey

20:00