Lecture 8 | Grammar of Graphics I

Max Pellert (https://mpellert.at)

IS 616: Large Scale Data Analysis and Visualization

Oxford English Dictionary, s.v. “grammar, n., sense 6.a”, July 2023. https://doi.org/10.1093/OED/2306046169

Why ggplot2?

https://github.com/ erikgahner/awesome-ggplot2

The transferrable skills from ggplot2 are not the idiosyncracies of plotting syntax, but a powerful way of thinking about visualisation, as a way of mapping between variables and the visual properties of geometric objects that you can perceive.

These ideas don’t come out of nowhere

Cox, D. R. (1978). Some Remarks on the Role in Statistics of Graphical Methods. Applied Statistics, 27(1), 4. https://doi.org/10.2307/2346220

built-in ggplot2
“beginner” “expert”
“basic” “advanced”
“easy” “hard”
“simple” “complicated”

ggplot2 built-in
“beginner” “expert”
“basic” “advanced”
“easy” “hard”
“simple” “complicated”

Pragmatic reasons

  • Functional data visualization

    1. Wrangle data
    2. Map data to visual elements
    3. Tweak scales, guides, axis, labels, theme

  • Easy to reason about how data drives visualization

  • Easy to iterate

  • Easy to be consistent

“This fits into a general principle I find myself arguing over and over, which is that you should teach your students as you would have wanted to be taught.”

http://varianceexplained.org/r/teach_ggplot2_to_beginners/

How do we express visuals in words?

“Good grammar is just the first step in creating a good sentence.”

What is a grammar of graphics?

  • Data to be visualized

  • Geometric objects that appear on the plot

  • Aesthetic mappings from data to visual component

  • Statistics transform data on the way to visualization

  • Coordinates organize location of geometric objects

  • Scales define the range of values for aesthetics

  • Facets group into subplots

gg is for “Grammar of Graphics”

Tidy Data

  1. Each variable forms a column

  2. Each observation forms a row

  3. Each observational unit forms a table

Start by asking

  1. What information do I want to use in my visualization?

  2. Is that data contained in one column/row for a given data point?

Data

ggplot(data)
country 1997 2002 2007
Canada 30.30584 31.90227 33.39014
China 1230.07500 1280.40000 1318.68310
United States 272.91176 287.67553 301.13995
country year pop
Canada 1997 30.30584
China 1997 1230.07500
United States 1997 272.91176
Canada 2002 31.90227

Data

Aesthetics

+ aes()

Map data to visual elements or parameters

  • year → x

  • pop → y

  • country → shape, color, etc.

Map data to visual elements or parameters

aes(
  x = year,
  y = pop,
  color = country
)

Data

Aesthetics

Geoms

+ geom_*()

Geometric objects displayed on the plot

See http://ggplot2.tidyverse.org/reference/ for many more options or just start typing geom_ in RStudio

 [1] "geom_abline"            "geom_area"              "geom_bar"              
 [4] "geom_bin_2d"            "geom_bin2d"             "geom_blank"            
 [7] "geom_boxplot"           "geom_col"               "geom_contour"          
[10] "geom_contour_filled"    "geom_count"             "geom_crossbar"         
[13] "geom_curve"             "geom_density"           "geom_density_2d"       
[16] "geom_density_2d_filled" "geom_density2d"         "geom_density2d_filled" 
[19] "geom_dotplot"           "geom_errorbar"          "geom_errorbarh"        
[22] "geom_freqpoly"          "geom_function"          "geom_hex"              
[25] "geom_histogram"         "geom_hline"             "geom_jitter"           
[28] "geom_label"             "geom_line"              "geom_linerange"        
[31] "geom_map"               "geom_path"              "geom_point"            
[34] "geom_pointrange"        "geom_polygon"           "geom_qq"               
[37] "geom_qq_line"           "geom_quantile"          "geom_raster"           
[40] "geom_rect"              "geom_ribbon"            "geom_rug"              
[43] "geom_segment"           "geom_sf"                "geom_sf_label"         
[46] "geom_sf_text"           "geom_smooth"            "geom_spoke"            
[49] "geom_step"              "geom_text"              "geom_tile"             
[52] "geom_violin"            "geom_vline"

Type Function
Point geom_point()
Line geom_line()
Bar geom_bar(), geom_col()
Histogram geom_histogram()
Regression geom_smooth()
Boxplot geom_boxplot()
Text geom_text()
Vert./Horiz. Line geom_{vh}line()
Count geom_count()
Density geom_density()

With programming, it’s OK first not to understand what you are doing

Load the libraries:

library(gapminder)
library(ggplot2)
library(gganimate)
library(gifski)
library(cowplot)

Inspect the data:

head(gapminder)
# A tibble: 6 × 6
  country     continent  year lifeExp      pop gdpPercap
  <fct>       <fct>     <int>   <dbl>    <int>     <dbl>
1 Afghanistan Asia       1952    28.8  8425333      779.
2 Afghanistan Asia       1957    30.3  9240934      821.
3 Afghanistan Asia       1962    32.0 10267083      853.
4 Afghanistan Asia       1967    34.0 11537966      836.
5 Afghanistan Asia       1972    36.1 13079460      740.
6 Afghanistan Asia       1977    38.4 14880372      786.

What about Python?

pip install plotnine

from plotnine import ggplot, geom_point, aes, stat_smooth, facet_wrap
from plotnine.data import mtcars

print(ggplot(mtcars, aes("wt", "mpg", color="factor(gear)"))
 + geom_point()
 + stat_smooth(method="lm")
 + facet_wrap("~gear"))

For a different summary of the data frame:

Rows: 1,704
Columns: 6
$ country   <fct> "Afghanistan", "Afghanistan", "Afghanistan", "Afghanistan", …
$ continent <fct> Asia, Asia, Asia, Asia, Asia, Asia, Asia, Asia, Asia, Asia, …
$ year      <int> 1952, 1957, 1962, 1967, 1972, 1977, 1982, 1987, 1992, 1997, …
$ lifeExp   <dbl> 28.801, 30.332, 31.997, 34.020, 36.088, 38.438, 39.854, 40.8…
$ pop       <int> 8425333, 9240934, 10267083, 11537966, 13079460, 14880372, 12…
$ gdpPercap <dbl> 779.4453, 820.8530, 853.1007, 836.1971, 739.9811, 786.1134, …

Let’s start with lifeExp vs gdpPercap

ggplot(gapminder) +
  aes(x = gdpPercap,
      y = lifeExp)

ggplot(gapminder) +
  aes(x = gdpPercap,
      y = lifeExp) +
  geom_point()

How can I tell countries apart?

ggplot(gapminder) +
  aes(x = gdpPercap,
      y = lifeExp,
      color = continent) +
  geom_point()

GDP is squished together on the left 

ggplot(gapminder) +
  aes(x = gdpPercap,
      y = lifeExp,
      color = continent) +
  geom_point() +
  scale_x_log10()

Still lots of overlap in the countries 

ggplot(gapminder) +
  aes(x = gdpPercap,
      y = lifeExp,
      color = continent) +
  geom_point() +
  scale_x_log10() +
  facet_wrap(~ continent) +
  guides(color = FALSE)

No need for color legend thanks to facet titles

Lots of overplotting due to point size 

ggplot(gapminder) +
  aes(x = gdpPercap,
      y = lifeExp,
      color = continent) +
  geom_point(size=0.25) +
  scale_x_log10() +
  facet_wrap(~ continent) +
  guides(color = FALSE)

Is there a trend? 

ggplot(gapminder) +
  aes(x = gdpPercap,
      y = lifeExp,
      color = continent) +
  geom_line() +
  geom_point(size=0.25) +
  scale_x_log10() +
  facet_wrap(~ continent) +
  guides(color = FALSE)

That line just connected all of the points sequentially… 

ggplot(gapminder) +
  aes(x = gdpPercap,
      y = lifeExp,
      color = continent) +
  geom_line(
    aes(group = country)
  ) +
  geom_point(size=0.25) +
  scale_x_log10() +
  facet_wrap(~ continent) +
  guides(color = FALSE)

💡We need time on the x-axis! 

ggplot(gapminder) +
  aes(x = year,
      y = gdpPercap,
      color = continent) +
  geom_line(
    aes(group = country)
  ) +
  geom_point(size=0.25) +
  scale_y_log10() +
  facet_wrap(~ continent) +
  guides(color = FALSE)

Can’t see x-axis labels, fix that 

ggplot(gapminder) +
  aes(x = year,
      y = gdpPercap,
      color = continent) +
  geom_point(size=0.25) +
  geom_line(
    aes(group = country)
  ) +
  scale_y_log10() +
  scale_x_continuous(
    breaks = seq(1950, 2000, 25)
  ) +
  facet_wrap(~ continent) +
  guides(color = FALSE)

What about life expectancy? 

ggplot(gapminder) +
  aes(x = year,
      y = lifeExp,
      color = continent) +
  geom_point(size=0.25) +
  geom_line(
    aes(group = country)
  ) +
  # scale_y_log10() +
  scale_x_continuous(
    breaks = seq(1950, 2000, 25)
  ) +
  facet_wrap(~ continent) +
  guides(color = FALSE)

Let’s add a trend line 

ggplot(gapminder) +
  aes(x = year,
      y = lifeExp,
      color = continent) +
  geom_line(
    aes(group = country)
  ) +
  geom_point(size=0.25) +
  geom_smooth() +
  scale_x_continuous(
    breaks = seq(1950, 2000, 25)
  ) +
  facet_wrap(~ continent) +
  guides(color = FALSE)

De-emphasize individual countries 

ggplot(gapminder) +
  aes(x = year,
      y = lifeExp,
      color = continent) +
  geom_line(
    aes(group = country),
    color = "grey75"
  ) +
  geom_point(size=0.25) +
  geom_smooth() +
  scale_x_continuous(
    breaks = seq(1950, 2000, 25)
  ) +
  facet_wrap(~ continent) +
  guides(color = FALSE)

Points are still in the way 

ggplot(gapminder) +
  aes(x = year,
      y = lifeExp,
      color = continent) +
  geom_line(
    aes(group = country),
    color = "grey75"
  ) +
  # geom_point(size=0.25) +
  geom_smooth() +
  scale_x_continuous(
    breaks = seq(1950, 2000, 25)
  ) +
  facet_wrap(~ continent) +
  guides(color = FALSE)

Let’s compare continents 

ggplot(gapminder) +
  aes(x = year,
      y = lifeExp,
      color = continent) +
  geom_line(
    aes(group = country),
    color = "grey75"
  ) +
  geom_smooth() +
  # scale_x_continuous(
  #   breaks = seq(1950, 2000, 25)
  # ) +
  # facet_wrap(~ continent) +
  guides(color = FALSE)

Wait, what color is each continent? 

ggplot(gapminder) +
  aes(x = year,
      y = lifeExp,
      color = continent) +
  geom_line(
    aes(group = country),
    color = "grey75"
  ) +
  geom_smooth() +
  theme(
  legend.position = "bottom"
  )

Let’s try the minimal theme 

ggplot(gapminder) +
  aes(x = year,
      y = lifeExp,
      color = continent) +
  geom_line(
    aes(group = country),
    color = "grey75"
  ) +
  geom_smooth() +
  theme_minimal() +
  theme(
  legend.position = "bottom"
  )

Fonts get cut off because they are too big 

ggplot(gapminder) +
  aes(x = year,
      y = lifeExp,
      color = continent) +
  geom_line(
    aes(group = country),
    color = "grey75"
  ) +
  geom_smooth() +
  theme_minimal(
    base_size = 8) +
  theme(
  legend.position = "bottom"
  )

Cool, but what about different population size? 

americas <- 
  gapminder %>% 
  filter(
    country %in% c(
      "United States",
      "Canada",
      "Mexico",
      "Ecuador"
    )
  )
# A tibble: 6 × 6
  country continent  year lifeExp      pop gdpPercap
  <fct>   <fct>     <int>   <dbl>    <int>     <dbl>
1 Canada  Americas   1952    68.8 14785584    11367.
2 Canada  Americas   1957    70.0 17010154    12490.
3 Canada  Americas   1962    71.3 18985849    13462.
4 Canada  Americas   1967    72.1 20819767    16077.
5 Canada  Americas   1972    72.9 22284500    18971.
6 Canada  Americas   1977    74.2 23796400    22091.

Let’s look at four countries in more detail.

How do their populations compare to each other? 

ggplot(americas) +
  aes(
    x = year,
    y = pop
  ) +
  geom_col()

But how many people are in each country?

ggplot(americas) +
  aes(
    x = year,
    y = pop,
    fill = country
  ) +
  geom_col()

Bars are “stacked”, how to separate them? 

ggplot(americas) +
  aes(
    x = year,
    y = pop,
    fill = country
  ) +
  geom_col(
    position = "dodge"
  )

position = "dodge" places objects next to each other instead of overlapping

🤓 What is scientific notation anyway? 

ggplot(americas) +
  aes(
    x = year,
    y = pop / 10^6,
    fill = country
  ) +
  geom_col(
    position = "dodge" 
  )

ggplot aesthetics can take expressions!

Might be easier to see countries individually 

ggplot(americas) +
  aes(
    x = year,
    y = pop / 10^6,
    fill = country
  ) +
  geom_col(
    position = "dodge" 
  ) +
  facet_wrap(~ country) +
  guides(fill = FALSE)

Let range of y-axis vary in each plot 

ggplot(americas) +
  aes(
    x = year,
    y = pop / 10^6,
    fill = country
  ) +
  geom_col(
    position = "dodge" 
  ) +
  facet_wrap(~ country,
             scales = "free_y") +
  guides(fill = FALSE)

Let’s pause and think how to combine the two parts of our analysis

To get inspiration, you can check out “The Best Stats You’ve Ever Seen” by Hans Rosling

http://www.ted.com/talks/ hans_rosling_shows_the_best_stats_you_ve_ever_seen

g_hr <- 
  ggplot(gapminder) +
  aes(x = gdpPercap, y = lifeExp, size = pop, color = country) +
  geom_point() +
  facet_wrap(~year)

g_hr <- 
  ggplot(gapminder) +
  aes(x = gdpPercap, y = lifeExp, size = pop, color = country) +
  geom_point() +
  facet_wrap(~year) +
  guides(color = FALSE, size = FALSE)

g_hr <- 
  g_hr +
  scale_x_log10(breaks = c(10^3, 10^4, 10^5),
                labels = c("1k", "10k", "100k")) +
  scale_color_manual(values = gapminder::country_colors) +
  scale_size(range = c(0.5, 12))

g_hr <- g_hr +
  labs(x = "GDP per capita", y = "Life Expectancy") +
  theme_minimal(base_family = "Fira Sans") +
  theme(strip.text = element_text(size = 16, face = "bold"),
    panel.border = element_rect(fill = NA, color = "grey40"),
    panel.grid.minor = element_blank())

ggplot(gapminder) +
  aes(x = gdpPercap, y = lifeExp, size = pop, color = country) +
  geom_point() +
  facet_wrap(~year) +
  guides(color = FALSE, size = FALSE) +
  scale_x_log10(
    breaks = c(10^3, 10^4, 10^5), 
    labels = c("1k", "10k", "100k")) +
  scale_color_manual(values = gapminder::country_colors) +
  scale_size(range = c(0.5, 12)) +
  labs(
    x = "GDP per capita",
    y = "Life Expectancy") +
  theme_minimal(14, base_family = "Fira Sans") +
  theme(
    strip.text = element_text(size = 16, face = "bold"),
    panel.border = element_rect(fill = NA, color = "grey40"),
    panel.grid.minor = element_blank())

Special Bonus: Animated!

https://gganimate.com/

# Same plot without facet_wrap()
g_hra <- 
  ggplot(gapminder) +
  aes(x = gdpPercap, y = lifeExp, size = pop, color = country) +
  geom_point() +
  guides(color = FALSE, size = FALSE) +
  scale_x_log10(
    breaks = c(10^3, 10^4, 10^5), 
    labels = c("1k", "10k", "100k")) +
  scale_color_manual(values = gapminder::country_colors) +
  scale_size(range = c(0.5, 12)) +
  labs(
    x = "GDP per capita",
    y = "Life Expectancy") +
  theme_minimal(18, base_family = "Fira Sans") +
  theme(
    plot.background = element_rect("#FAFAFA", color = NA),
    strip.text = element_text(size = 16, face = "bold"),
    panel.border = element_rect(fill = NA, color = "grey40"),
    panel.grid.minor = element_blank(),
    plot.title = element_text(hjust = 0.5)
  ) + 
  transition_states(
    year, 1, 0
  ) + 
  ggtitle("{closest_state}")

animate(g_hra, duration = 10, fps = 24, width = 700, height = 500, renderer = gifski_renderer())
anim_save("hans-rosling-esque.gif")

Acknowledgements

http://github.com/gadenbuie/gentle-ggplot2