Deep dive into ggplot2 layers - I

Lecture 1

Dr. Greg Chism

University of Arizona
INFO 526 - Fall 2024

Announcements

  • A note on readings for this week: some of it is review so feel free to skim those parts.
  • HW 0 is due Monday (Sept 02) - This is ungraded, but serves as good practice
  • RQ 1 is due Wednesday (Sept 04)

Setup

# load packages
library(tidyverse)
library(here)
library(countdown)

# set theme for ggplot2
ggplot2::theme_set(ggplot2::theme_minimal(base_size = 14))

# set width of code output
options(width = 65)

# set figure parameters for knitr
knitr::opts_chunk$set(
  fig.width = 7,        # 7" width
  fig.asp = 0.618,      # the golden ratio
  fig.retina = 3,       # dpi multiplier for displaying HTML output on retina
  fig.align = "center", # center align figures
  dpi = 300             # higher dpi, sharper image
)

A/B testing

Data: Sale prices of houses in Tucson

  • Data on houses for sale
    in Tucson, AZ, around July 2023

  • Scraped from Zillow

  • Source: tucsonHousing.csv

Modernist house in Tucson AZ

slides/data/tucsonHousing.csv

library(tidyverse)
library(here)

tucsonHousing <- read_csv(here(
  "slides", "01", "data" ,"tucsonHousing.csv"))

glimpse(tucsonHousing)
Rows: 112
Columns: 8
$ address    <chr> "710 E 5th St, Tucson, AZ 85719", "3543 N Fl…
$ year_built <dbl> 1936, 1943, 1948, 1950, 1950, 1951, 1951, 19…
$ price      <dbl> 330000, 260000, 310000, 270000, 270000, 2149…
$ bed        <dbl> 2, 2, 3, 4, 4, 3, 1, 4, 3, 3, 4, 4, 3, 3, 4,…
$ bath       <dbl> 1, 1, 1, 2, 2, 1, 1, 2, 1, 2, 2, 2, 1, 2, 2,…
$ area       <dbl> 903, 1253, 1256, 1634, 1634, 1070, 766, 1490…
$ type       <chr> "Single Family", "Single Family", "Single Fa…
$ url        <chr> "/homedetails/710-E-5th-St-Tucson-AZ-85719/8…

A simple visualization

ggplot(tucsonHousing, aes(x = area, y = price)) +
  geom_point(alpha = 0.7, size = 2) +
  geom_smooth(method = "lm", se = FALSE, size = 0.7) +
  labs(
    x = "Area (square feet)",
    y = "Sale price (USD)",
    title = "Price and area of houses in Tucson"
  )

New variable: decade_built

tucsonHousing <- tucsonHousing |>
  mutate(decade_built = (year_built %/% 10) * 10)

tucsonHousing |>
  select(year_built, decade_built)
# A tibble: 112 × 2
   year_built decade_built
        <dbl>        <dbl>
 1       1936         1930
 2       1943         1940
 3       1948         1940
 4       1950         1950
 5       1950         1950
 6       1951         1950
 7       1951         1950
 8       1952         1950
 9       1952         1950
10       1952         1950
# ℹ 102 more rows

New variable: decade_built_cat

tucsonHousing <- tucsonHousing |>
  mutate(
    decade_built_cat = case_when(
      decade_built <= 1950 ~ "1950 or before",
      decade_built >= 2000 ~ "2000 or after",
      TRUE ~ as.character(decade_built)
    )
  )

tucsonHousing |>
  count(decade_built_cat)
# A tibble: 6 × 2
  decade_built_cat     n
  <chr>            <int>
1 1950 or before      25
2 1960                 9
3 1970                18
4 1980                 9
5 1990                20
6 2000 or after       31

A slightly more complex visualization

ggplot(
  tucsonHousing,
  aes(x = area, y = price, color = decade_built_cat)
) +
  geom_point(alpha = 0.7, show.legend = FALSE) +
  geom_smooth(method = "lm", se = FALSE, size = 0.5, show.legend = FALSE) +
  facet_wrap(~decade_built_cat) +
  labs(
    x = "Area (square feet)",
    y = "Sale price (USD)",
    color = "Decade built",
    title = "Price and area of houses in Tucson"
  )

A/B testing

In the next two slides, the same plots are created with different “cosmetic” choices. Examine the plots two given (Plot A and Plot B), and indicate your preference by voting for one of them in the Vote tab.

Test 1

Test 2

What makes figures bad?

Bad taste

Data-to-ink ratio

Tufte strongly recommends maximizing the data-to-ink ratio this in the Visual Display of Quantitative Information (Tufte, 1983).

Graphical excellence is the well-designed presentation of interesting data—a matter of substance, of statistics, and of design … [It] consists of complex ideas communicated with clarity, precision, and efficiency. … [It] is that which gives to the viewer the greatest number of ideas in the shortest time with the least ink in the smallest space … [It] is nearly always multivariate … And graphical excellence requires telling the truth about the data. (Tufte, 1983, p. 51).

Cover of The Visual Display of Quantitative Information

Which of the plots has higher data-to-ink ratio?

A deeper look

at the plotting code

Summary statistics

mean_area_decade <- tucsonHousing |>
  group_by(decade_built_cat) |>
  summarise(mean_area = mean(area))

mean_area_decade
# A tibble: 6 × 2
  decade_built_cat mean_area
  <chr>                <dbl>
1 1950 or before       1440.
2 1960                 1506.
3 1970                 1558.
4 1980                 1586.
5 1990                 1570.
6 2000 or after        1795.

Barplot

ggplot(
  mean_area_decade,
  aes(y = decade_built_cat, x = mean_area)
) +
  geom_col() +
  labs(
    x = "Mean area (square feet)", y = "Decade built",
    title = "Mean area of houses in Tucson, by decade built"
  )

Scatterplot

ggplot(
  mean_area_decade,
  aes(y = decade_built_cat, x = mean_area)
) +
  geom_point(size = 4) +
  labs(
    x = "Mean area (square feet)", y = "Decade built",
    title = "Mean area of houses in Tucson, by decade built"
  )

Lollipop chart – a happy medium?

Bad data

Faceted plot showing the average importance of democracy in 6 countries over time.

Faceted plot showing the average importance of democracy in 6 countries over time.

Bad perception

Aspect ratios affect our perception of rates of change, modeled after an example by William S. Cleveland.

Aesthetic mappings in ggplot2

A second look: lollipop chart

ggplot(
  mean_area_decade,
  aes(y = decade_built_cat, x = mean_area)
) +
  geom_point(size = 4) +
  geom_segment(aes(
    x = 0, xend = mean_area,
    y = decade_built_cat, yend = decade_built_cat
  )) +
  labs(
    x = "Mean area (square feet)",
    y = "Decade built",
    title = "Mean area of houses in Tucson, by decade built"
  )

Activity: Spot the differences |

ggplot(
  mean_area_decade,
  aes(y = decade_built_cat, x = mean_area)
) +
  geom_point(size = 4) +
  geom_segment(aes(
    xend = 0,
    yend = decade_built_cat
  )) +
  labs(
    x = "Mean area (square feet)",
    y = "Decade built",
    title = "Mean area of houses in Tucson, by decade built"
  )

Can you spot the differences between the code here and the one provided in the previous slide? Are there any differences in the resulting plot? Work in a pair (or group) to answer.

03:00

Global vs. layer-specific aesthetics

  • Aesthetic mappings can be supplied in the initial ggplot() call, in individual layers, or in some combination of both.

  • Within each layer, you can add, override, or remove mappings.

  • If you only have one layer in the plot, the way you specify aesthetics doesn’t make any difference. However, the distinction is important when you start adding additional layers.

Activity: Spot the differences II

Do you expect the following plots to be the same or different? If different, how? Discuss in a pair (or group) without running the code and sketch the resulting plots based on what you think the code will produce.

Plot

# Plot A
ggplot(tucsonHousing, aes(x = area, y = price)) +
  geom_point(aes(color = decade_built_cat))
# Plot B
ggplot(tucsonHousing, aes(x = area, y = price)) +
  geom_point(color = "blue")
# Plot C
ggplot(tucsonHousing, aes(x = area, y = price)) +
  geom_point(color = "#a493ba")
03:00

Wrap up

Think back to all the plots you saw in the lecture, without flipping back through the slides. Which plot first comes to mind? Describe it in words.

🔗 datavizaz.org