Grammar of Graphics

Grammar of Graphics

Source: ggplot2 workshop by @thomasp85

Grammar of Graphics

Source: ggplot2 workshop by @thomasp85

• Your dataset

• Tidy format

Grammar of Graphics

Source: ggplot2 workshop by @thomasp85

• This is how we tell R which variables we want to plot

• Aesthetics mapping
  Links variable in the data to graphical properties

• Facets mapping
  Links variable in data to panels in the plot layout

Grammar of Graphics

Source: ggplot2 workshop by @thomasp85

• Even tidy data may need some transformation

• Transform input variables to displayed values

  • Bins for histogram
  • Summary statistics for boxplot
  • No. of observations in a category for bar chart

• Implicit in many plot types

Grammar of Graphics

Source: ggplot2 workshop by @thomasp85

• Help you interpret the plot

  • Categories -> color
  • Numeric -> position

• Automatically generated in ggplot and can be customized

  • log scale
  • time series

Grammar of Graphics

Source: ggplot2 workshop by @thomasp85

• Aesthetics as graphical repersentations

• Determines your plot type

  • bar chart
  • scatter
  • boxplot
  • ...

Grammar of Graphics

Source: ggplot2 workshop by @thomasp85

• Divide your data into panels using one or two groups

• Allows you to look at smaller subsets of data

Grammar of Graphics

Source: ggplot2 workshop by @thomasp85

• Positions are interpreted by the coordinate system

• Defines the physical mapping of the aesthetics

Grammar of Graphics

Source: ggplot2 workshop by @thomasp85

• Overall look of the plot

• Spans every part of the graphic that is not linked to the data

  • "non-data ink"

Getting Started

• Load the tidyverse package

library(tidyverse)

• If this is your first time you may have to install it first

install.packages("tidyverse")
library(tidyverse)

Your first ggplot

ggplot(data=mpg)+
  aes(x=displ)+
  aes(y=hwy)+
  geom_point()

What did we need?

Source: ggplot2 workshop by @thomasp85

Common Problems

• Make sure that every ( is matched with a )

• Make sure that every " is paired with another "

• Make sure that + is in the right place: it has to come at the end of the line, not the start. The following code will not work

ggplot(data = mpg) 
+ geom_point(mapping = aes(x = displ, y = hwy))

• Look for help by typing ?function_name

  • scroll down to examples

• Look at the error message

  • try googling the error message

Aesthetics

ggplot(data=mpg)+
  aes(x=displ)+
  aes(y=hwy)+
  aes(shape=class)+
  geom_point()

Aesthetics

Setting the properties of geom manually

ggplot(data = mpg) + 
  geom_point(mapping = aes(x = displ, y = hwy), color = "blue")

Here, the color "blue" doesn’t convey information about a variable, but only changes the appearance of the plot

Aesthetics

To set a geometric property manually, place it outside of aes()

• The name of a color as a character string

• The size of a point in mm

• The shape of a point as a number

R has 25 built in shapes that are identified by numbers

Aesthetics

Remember aesthetics depend on geometry...

Mulitple geoms

ggplot(data = mpg,
       mapping = aes(
         x = displ,
         y = hwy)) +
  geom_point(
    mapping = aes(
      color = class)) +
  geom_smooth()

Mulitple geoms

ggplot(data = mpg,
       mapping = aes(
         x = displ,
         y = hwy)) +
  geom_point(
    mapping = aes(
      color = class)) +
  geom_smooth(data =
                filter(mpg,
                       class == "suv"),
              se = FALSE)

Statistical Transformations

Statistical Transformations

ggplot(data = diamonds) + 
  geom_bar(mapping = aes(x = cut))

Where does count on y-axis come from?

Statistical Transformations

The algorithm used to calculate new values for a graph is called a stat, short for statistical transformation

Statistical Transformations

You can find out which stat each geom uses by looking at the default value of the stat argument of the help page.

What it the default stat for geom_bar?

Statistical Transformations

• Overriding default options
• Here, display bar chart of proportions instead of count

ggplot(data = diamonds) + 
  geom_bar(mapping = aes(x = cut, y = stat(prop),
                         group = 1))

Position Adjustments

ggplot(data = diamonds) + 
  geom_bar(mapping = aes(x = cut, 
                         colour = cut))

Position Adjustments

ggplot(data = diamonds) + 
  geom_bar(mapping = aes(x = cut, 
                         fill = cut))

Position Adjustments

ggplot(data = diamonds) + 
  geom_bar(mapping = aes(x = cut, 
                         fill = clarity))

position="identity"

• places each object exactly where it falls in the context of the graph
• useful if bars are made transparent

ggplot(data = diamonds, mapping = aes(x = cut, colour = clarity)) + 
  geom_bar(fill = NA,
           position = "identity")

position="fill"

• makes each set of stacked bars the same height
• Useful for comparing proportions across groups

ggplot(data = diamonds) + 
  geom_bar(mapping = aes(x = cut, fill = clarity), 
           position = "fill")

position="dodge"

• Places objects next to each other
• Useful for comparing individual values

ggplot(data = diamonds) + 
  geom_bar(mapping = aes(x = cut, fill = clarity), 
           position = "dodge")

Scales

Source: ggplot2 workshop by @thomasp85

Scales

ggplot(mpg) +
  geom_point(aes(x = displ, y = hwy, colour = class))

Scales

ggplot(mpg) +
  geom_point(aes(x = displ, y = hwy, colour = class))+
  scale_colour_brewer(type = 'qual')

Scales

ggplot(mpg) +
  geom_point(aes(x = displ, y = hwy)) +
  scale_x_continuous(breaks = c(3, 5, 6)) +
  scale_y_continuous(trans = 'log10')

Facets

ggplot(data = mpg) +
  geom_point(mapping = aes(
    x = displ,
    y = hwy)) +
  facet_grid(drv ~ cyl)

Coordinate Systems

ggplot(data = mpg,
       mapping = aes(
         x = class,
         y = hwy)) +
  geom_point(position = "jitter") +
  coord_polar()

Themes

ggplot(data=mpg)+
  aes(x=displ)+
  aes(y=hwy)+
  geom_point()+
  theme_dark()

GGPLOT object

myplot <- ggplot(data=mpg)+
  aes(x=displ)+
  aes(y=hwy)+
  geom_point()+
  theme_minimal()
myplot+
  labs(x="Displacement")+
  labs(y="Highway Mileage")+
  labs(title="My first GGPLOT")+
  labs(subtitle="This is the subtitle")+
  labs(caption="Source:mpg dataset")

A ggplot template

ggplot(data = <DATA>) +
<GEOM_FUNCTION>(
mapping = aes(<MAPPINGS>),
stat = <STAT>,
position = <POSITION>
) +
<COORDINATE_FUNCTION> +
<FACET_FUNCTION>

In practice, you rarely need to supply all seven parameters to make a graph because ggplot2 will provide useful defaults for everything except the data, the mappings, and the geom function.

The layered grammar of graphics

R for Data Science by Hadley WickHam

The layered grammar of graphics

R for Data Science by Hadley WickHam

The layered grammar of graphics

R for Data Science by Hadley WickHam

Plot Composition

install.packages("patchwork") 
library(patchwork)

Plot Composition

library(ggplot2)
library(patchwork)
p1 <- ggplot(mpg) + geom_point(aes(displ, hwy)) # first plot
p2 <- ggplot(mpg) + geom_boxplot(aes(displ, hwy, group = class)) # second plot 
p1+p2 # combined plot output using patchwork package

Plot Composition

p3 <- ggplot(mpg, aes(displ, hwy))+geom_point(aes(color=class))+geom_smooth(aes(color=class))
p4 <- ggplot(mpg) + geom_bar(aes(class))
(p1 | p2 | p3) /
      p4

Plot Annotation

• can add annotations by code

• packages ggrepel and ggforce

Plot Annotation

ggplot(mpg[1:20,], aes(x = displ, y = hwy)) +
  geom_point() +
  geom_text(aes(label = model))

Plot Annotation

library(ggrepel)
ggplot(mpg[1:20,], aes(x = displ, y = hwy)) +
  geom_point() +
  geom_text_repel(aes(label = model))

Plot Annotation

library(ggforce)
ggplot(mpg, aes(x = displ, y = hwy)) +
  geom_point()+
  geom_mark_ellipse(
    aes(filter = class  == "2seater",
        label = '"2 seater"/Sports Cars',
        description = 'Sports cars have large engines but small bodies, which improves their mileage'))

What next?

• ggplot2 extensions
• Rstudio cheatsheet
• BBC Visual and Data Journalism cookbook for R graphics

More Insipiration

• Tidy Tuesday Project

• "#TidyTuesday" on Twitter

• Youtube screencasts by David Robinson

Resources Used

• flipbookr package by Gina Reynolds

• xaringan package by Yihui Xie

• R for Data Science book by Hadley Wickham & Garrett Grolemund

• ggplot2 workshop by Thomas Lin Pedersen

• Illustrations by Allison Horst