10 min read

An Introduction to ggplot2 (Workshop)

2022-05-17

This is an adaptation of a workshop I ran last year as part of the Centre of Excellence for the Dynamics of Language (ANU) Seminar Series- the main change is that I’ve subbed out the data with the iris data that’s automatically available in R so that it’s more accessible. If you would like a recording of the workshop, which uses vowel data, you can email me at

Workshop Overview

  • Introduction to the ggplot2 package and its main functions
  • How to make three plots
    • Scatter plot
    • Box plot
    • Density plot
  • How to customise these plots
    • Colour
    • Labels
    • Legends
    • Themes
    • Faceting
    • Manual palettes

The Basics

  • ggplot2 is a R package dedicated to data visualisation

  • Has an underlying grammar that allows you to build graphs by combining independent components

    • This allows you to build almost any type of chart
    • The same underlying data can be transformed by many different scales or layers (i.e., it is extremely flexible)

  • It is also over a decade old, meaning there are a lot of resources available

  • To use ggplot2, you need to:

    • Install and load either the ggplot2 package or the tidyverse package
    • Load the data (name_of_object <- read.csv("name_of_spreadsheet.csv"))

  • I have also included code for installing and loading the RColorBrewer package

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

The base layer

  • All plots are composed of:
    • The data: the information you want to visualise
    • A mapping: a description of how you want the variables in your data to be ‘mapped’ to aesthetic attributes like colour, shapes, or x and y axes
  • All plots you make in ggplot2 will begin with the ggplot() function
    • This builds the first component of your graph (the base layer)
  • You also need to tell ggplot what data you want to visualise
    • The name of your dataframe or object, in our case ‘cars’
  • The code below will create an empty base layer
ggplot(data = iris)

Mapping

  • Mapping depends on what kind of graph you are after, but for most you will want to add x and/or y axes
  • Within ggplot(data=ban), we need another function aes() within which you give the x and y axes so that:
    • ggplot(data=*dataframe*, aes(x=*column_a*, y=*column_b*)) 
    • The names you give for ‘x’ and ‘y’ are the names of the columns in the dataframe you want to plot
  • The code below will create a graph with a labeled x and y axis that is otherwise empty. In the next section, we will turn it into a scatter plot
ggplot(data = iris, aes(x = Sepal.Length, y = Petal.Length))

Three Basic Plots

Basic scatter plot

  • We already have the basis for a scatterplot from the previous code as we have two continuous variables on the x and y axes
  • To turn this into a scatterplot, add + geom_point() to the previous code
    • This tells ggplot2 that we want to build a scatter plot, with the specified x and y axes
  • When you add a new component to a graph, there must be a + connecting to the previous one
  • And voila!
ggplot(data = iris, aes(x = Sepal.Length, y = Petal.Length)) +
  geom_point()

Basic boxplot

  • The basic ggplot(data=, aes(x=,y=)) stays the same
    • Instead of +geom_point() we add +geom_boxplot()
  • Box plots use:
    • a categorical variable on the x axis
    • a continuous variable on the y axis
ggplot(data = iris, aes(x = Species, y = Petal.Length)) + 
  geom_boxplot()

Basic density plot

  • The basic code is ggplot(data=, aes(x=))
    • This time we add +geom_density()
  • Density plots use:
    • a single continuous variable on the x-axis
    • Y axis tells you the distribution of this variables
ggplot(data = iris, aes(x = Petal.Length)) +
  geom_density()

Practice Mini-Tasks

1: Build a scatter plot

  • Make your own scatter plot by changing the x and y axes to different continuous variables from the data:
    • Petal.Length
    • Petal.Width
    • Sepal.Length
    • Sepal.Width

2: Build a box plot

  • Make your own box plot by changing the y axis to different continuous variables from the dataframe:
    • Petal.Length
    • Petal.Width
    • Sepal.Length
    • Sepal.Width
  • And change the x axis to a different categorical variable from the dataframe (although with this data there is only one option)
    • Species

3: Build a density plot

  • Make your own density plot by changing the x axis to a different continuous variable from the dataframe
    • Petal.Length
    • Petal.Width
    • Sepal.Length
    • Sepal.Width

Customising in ggplot2

Colour

  • Colour is a very easy way to add additional information
  • We map colour to variables within the aes() function, after we have put in the x and y axes
    • ggplot(data=data_frame, aes(x=columm_a,y=column_b,color=column_c))
  • The same principle applies to shape and linetype
ggplot(data = iris, aes(x = Sepal.Length, y = Petal.Length, 
                        color = Species)) +
  geom_point()

  • If your graph has shapes with lines (like boxplots and density plots), you can:
    • change the line colour with color=column_c
    • change the colour inside the lines with fill=column_d
ggplot(data = iris, aes(x = Sepal.Length, color = Species))+
  geom_density()

ggplot(data = iris, aes(x = Sepal.Length, fill = Species)) +
  geom_density()

### Labels and Titles

  • The +labs() function lets you change the plot title, caption, x and y axes and the legend labels
    • labs(title = "Plot Title", caption = "Plot Caption", x = "column_a", y = "column_b", color = "column_c", fill = "column_d")
  • To change the legend labels, you specify the mapping attribute (color or fill), then the column being mapped to that attribute
ggplot(data = iris, aes(x = Sepal.Length, y = Petal.Length, 
                        color = Species)) +
  geom_point() +
  labs(title = "Sepal Length and Petal Length",
      caption = "Data from R", 
      x = "Sepal Length", y = "Petal Length",
      color = "Species")

Themes

  • You can change the background of the chart
    • Grey squares is the default
  • Other options include:
    • +theme_bw(), +theme_light(), +theme_dark(), +theme_classic()
ggplot(data = iris, aes(x = Sepal.Length, y = Petal.Length,
                        color = Species)) +
  geom_point() +
  labs(title = "Sepal Length and Petal Length",
      caption = "Data from R", 
      x = "Sepal Length", y = "Petal Length",
      color = "Species")+
  theme_bw()

Legends

  • The default legend position is to the right of the plot
  • It can be changed using +theme(legend.position="")
    • Options are “bottom”, “top”, “left”, or “right”
  • You can also alter the legend name and labels using:
  • scale_color_discrete(name="",labels=c("","")) and
  • scale_fill_discrete(name="",labels=c("",""))
  • You can remove one of the legends if you have more than one with guides(fill="none") or guides(color="none")
    • Need to specify the mapping you want to remove
    • If you want to remove all legends you can use + theme(legend.position="none")
ggplot(data = iris, aes(x = Sepal.Length, y = Petal.Length, color = Species)) +
  geom_point() +
  labs(title = "Sepal Length and Petal Length",
      caption = "Data from R", 
      x = "Sepal Length",y = "Petal Length",
      color = "Species") +
  theme_bw() +
  theme(legend.position = "left")

Mini-tasks: Customising your chart

1: Colour

  • Customise a scatter plot, box plot and density plot with colour
  • Scatter plot
    • ggplot(data = data_frame, aes(x = column_1, y = column_2, color = column_3)) + geom_point()
  • Box plot
    • ggplot(data = ban, aes(x = age, y = F1_lob, color = gender, fill = community)) + geom_boxplot()
  • Density plot
    • ggplot(data = ban, aes(x = F1_lob, color = age, fill = community)) + geom_density()

2: Customising with colour

  • Customise a scatter plot, box plot and density plot with colour
  • Scatter plot
    • ggplot(data = data_frame, aes(x = column_1, y = column_2, color = column_3)) + geom_point()
  • Box plot
    • ggplot(data = ban, aes(x = age, y = F1_lob, color = gender, fill = community)) + geom_boxplot()
  • Density plot
    • ggplot(data = ban, aes(x = F1_lob, color = age, fill = community)) + geom_density()

3: Labels and Titles

  • Give your plot from Task 4 a new title!
  • Try and change the labels of the x and y axes too
    • Previous plot code +labs(title = "Plot Title", caption = "Plot Caption", x = "column_a", y = "column_b", color = "column_c", fill = "column_d")

4: Theme

  • Change the theme for the plot from Task 5
    • +theme_bw()
    • +theme_light()
    • +theme_dark()
    • +theme_classic()

5: Legends

  • Change the labels for your legend
  • Try and remove one of them if you have two

Faceting

facet_wrap()

  • You can break down your graph further by categorical variables with facet_wrap(), which automatically wraps graphs in a rectangle layout
  • facet_wrap(~) can take one or two arguments
    • for one argument, it goes to the right of the ~ facet_wrap(~ column_e)
    • for two arguments, one goes on either side of the ~ facet_wrap(column_e ~ column_f)
ggplot(data = iris, aes(x = Sepal.Length, y = Petal.Length)) +
  geom_point() +
  labs(title = "Sepal Length and Petal Length",
      caption = "Data from R", 
      x = "Sepal Length", y = "Petal Length",
      color = "Species") +
  theme_bw() +
  facet_wrap(~ Species)

facet_grid()

  • facet_grid() can facet into columns or rows, or both
    • facet_grid(cols = vars(column_e)) will facet into columns based on this variable
    • facet_grid(rows = vars(column_f)) will facet into rows based on this variable
    • facet_grid(cols = vars(column_e), rows = vars(column_f)) will facet into rows and columns based on the two variables
ggplot(data = iris, aes(x = Sepal.Length, y = Petal.Length))+
  geom_point() +
  labs(title = "Sepal Length and Petal Length",
      caption = "Data from R", 
      x = "Sepal Length", y = "Petal Length",
      color = "Species") +
  theme_bw() +
  facet_grid(cols = vars(Species))

Changing the colours

  • When you map variables onto colour, R will automatically select colours for you
    • But often it’s not cute or doesn’t match the colour scheme of your presentation
  • ggplot lets us change the colours. You can:
    • Manually select each colour within the plot
    • Choose a pre-existing palette

Manually selecting colours

  • Get a list of the available colour names with colors()
  • Or you can use the Hex code for the colour
  • To change the colours in a plot manually we add scale_color_manual()
    • Inside the () we add values=c() with the colour names or hex codes in “”
    • When using hex codes, remember to have the # in front of the code
ggplot(data = iris, aes(x = Sepal.Length, y = Petal.Length, color = Species)) +
  geom_point() +
  scale_color_manual(values = c("blue2", "chartreuse2", "blueviolet"))

  • Where you have both fill and colour, the same applies
  • To change the fill colour manually add:
    • +scale_fill_manual(values = c())
  • To change the line colour manually add:
    • +scale_color_manual(values = c())

Pre-existing palettes

  • You don’t need to go through choosing colours manually if you don’t want to
  • There are a lot of pre-existing palettes that you can add to your graph
  • Different packages have different palettes, I use RColorBrewer
  • We replace the ‘manual’ in the previous code with brewer:
    • scale_color_brewer(palette = "")
    • scale_fill_brewer(palette = "")
  • You can find palette names in the link in the script
ggplot(data = iris, aes(x = Sepal.Length, y = Petal.Length, color = Species)) +
  geom_point() +
  scale_color_brewer(palette = "RdGy", 
          name = "Iris Species")

Mini-tasks: Make it cute and bring it together

1: Make it cute

  • Manually change the colours in a chart by changing the colours manually, or using a pre-existing palette _ You can copy and paste a graph from above!

2: Bring it together

  • Construct a box plot, density plot or scatter plot (i.e. copy and paste from different tasks)
    • Change the theme
    • Change the title and labels (axes and legend)
    • Change the legend (presence or position)
    • Facet it by one or two categorical variables
    • Customise the colours manually or with a pre-existing palette

Troubleshooting

  • Remember to have colour and palette names/hex codes in ” ”
  • “” and () have to be closed
  • R is case-sensitive and space-sensitive
  • there must be a + connecting lines
  • Hex codes must have the # before the digits