day6pm-slides

EDS 221 Day 6 PM

Grammar of graphics

August 17^th, 2026

The grammar of graphics

ggplot2 builds a figure by layering independent pieces on top of each other, instead of picking one big “chart type.”

data

The data frame you’re plotting.

mapping

Which columns control which visual properties (aes()).

geometry

The shape used to represent the data (geom_*()).

scale

How the mapping shows up on the figure (scale_*()).

Geometries: start with data

ggplot() sets up a plot for a data frame — with nothing mapped yet, there’s nothing to draw.

library(tidyverse)
library(palmerpenguins)

ggplot(data = penguins)

Geometries: add a mapping

aes() maps columns to axes. This sets up the grid, but still no geometry has been added.

ggplot(
  data = penguins,
  mapping = aes(x = flipper_length_mm, y = body_mass_g)
)

Geometries: add a geometry

A geom_*() layer draws the mapped data. geom_point() draws one point per row.

ggplot(
  data = penguins,
  mapping = aes(x = flipper_length_mm, y = body_mass_g)
) +
  geom_point()

Geometries: choosing a geometry

The geometry you choose depends on the kind of relationship you want to show.

One numeric variable → geom_histogram()

ggplot(penguins, aes(x = body_mass_g)) +
  geom_histogram()

One categorical variable → geom_bar()

ggplot(penguins, aes(x = island)) +
  geom_bar()

Geometries: comparing groups

A numeric variable by a categorical variable → geom_boxplot() shows the distribution within each group.

ggplot(
  data = penguins,
  mapping = aes(x = species, y = body_mass_g)
) +
  geom_boxplot()

Aesthetics

aes() isn’t just for x and y — it can map columns to other visual properties too.

x, y — position
color — outline / line color
fill — interior color
shape — point symbol

Aesthetics: color vs. fill

color — outlines

ggplot(
  penguins,
  aes(x = species, y = body_mass_g, color = sex)
) +
  geom_boxplot()

fill — interiors

ggplot(
  penguins,
  aes(x = species, y = body_mass_g, fill = sex)
) +
  geom_boxplot()

Aesthetics: shape

shape gives each category of a point geometry a different symbol — useful when color alone isn’t enough (e.g., grayscale printing).

ggplot(
  data = penguins,
  mapping = aes(x = body_mass_g, y = flipper_length_mm, shape = species)
) +
  geom_point()

Aesthetics: combining color and shape

Mapping species to both shape and color encodes the same grouping two ways, making it easier to read.

ggplot(
  data = penguins,
  mapping = aes(
    x = body_mass_g,
    y = flipper_length_mm,
    shape = species,
    color = species
  )
) +
  geom_point()

Scales

Every mapped aesthetic has a scale controlling how data values become visuals — an axis, a legend, a color ramp. scale_*() functions let you customize them: titles, breaks, and the values themselves.

Scales: renaming and rebreaking axes

scale_x_continuous() and scale_y_continuous() control the x- and y-axis scales — here, a cleaner title and fewer tick marks (n.breaks).

ggplot(
  data = penguins,
  mapping = aes(x = body_mass_g, y = flipper_length_mm, shape = species)
) +
  geom_point() +
  scale_x_continuous(name = "Body mass (g)", n.breaks = 2) +
  scale_y_continuous(name = "Flipper length (mm)", n.breaks = 4)

Scales: choosing values manually

scale_shape_manual() lets you assign the exact shape (or color, or fill) for each category, instead of accepting the default.

ggplot(
  data = penguins,
  mapping = aes(x = body_mass_g, y = flipper_length_mm, shape = species)
) +
  geom_point() +
  scale_shape_manual(
    name = "Species",
    values = c(Adelie = "square", Chinstrap = "diamond", Gentoo = "circle")
  )