Session details


  1. To become aware of the powerful features of ggplot2.
  2. To learn about some of the fundamentals of easily creating amazing graphics.
  3. To know about some resources to continue learning.

At the end of this session, you will achieve this objective by creating a fairly simple, visually-appealing graph that shows:

Resources for learning and help

For learning:

For help:

Quickly get familiar with data to plot

For this session we will be using the CO2 dataset. Here is some code to get a sense of the data.

# Variables

# General contents

# Quick summary

Exercise: Choose a dataset and check it out

There are several exercises in this session. Choose one of the below datasets and use that dataset for all later exercises.

For complete R beginners, use:

For more confident R users, use one of these:

Check out the contents of the dataset you choose using:

# variable names of dataset

# contents of dataset

# summary of dataset

Basic structure of using ggplot2

ggplot2 uses the "Grammar of Graphics" (gg). This is a powerful approach to creating plots because it provides a consistent way of telling ggplot2 what to do. There are at least three aspects to using ggplot2 that relate to the grammar:

To maximise the power of ggplot2, make heavy use of autocompletion. You can do this by typing, for instance, geom_ and then hitting the TAB key to see a list of all the geoms. Or after typing theme(, hit TAB to see all the options inside theme.

Visualise 1-dimensional (x axis) data

There are many ways of showing plotting continuous (e.g. weight, height) variables in ggplot2. For discrete (e.g. terrain type: mountain, plains, or sex: woman, man) variables, there is really only one way.


# Continuous
ggplot(CO2, aes(x = conc)) +

# Discrete
ggplot(CO2, aes(x = Treatment)) +

Exercise: One variable plots

Time: 10 min

# put name of dataset below

# use dataset with one continuous variable
ggplot(___, aes(x = ___)) +
    # finish the geom to create either a histogram, freqpoly, or density layer

# use dataset with one discrete variable
ggplot(___, aes(x = ___)) +
    # finish the geom to create a bar layer

Visualise 2-dimensional (x and y axis) data

You can of course include data on the y axis too! This is usually what you use graphs for! There are many more types of "geoms" to use for having data on both axes, and which one you choose depends on what you are trying to show and what the data is like. Usually you put the variable that you can influence (the independent variable) on the x axis and the variable that responds (the dependent variable) on the y axis.

# Using continuous data
co2_plot_nums <- ggplot(CO2, aes(x = conc, y = uptake))

# Standard scatter plot
co2_plot_nums + geom_point()

# Connect all the data with a line
co2_plot_nums + geom_line()

# Put overlapping data into "hexes".. useful for massive datasets
co2_plot_nums + geom_hex()

# Connects data as they appear in the dataset
co2_plot_nums + geom_path()

# Runs a smoothing line with confidence interval
co2_plot_nums + geom_smooth()

# Using mixed data
co2_plot_mixed <- ggplot(CO2, aes(x = Type, y = uptake))

# Standard boxplot
co2_plot_mixed + geom_boxplot()

# Bar plot, showing total sum of uptake
co2_plot_mixed + geom_col()

# Better than boxplot, show the actual data!
co2_plot_mixed + geom_jitter()

# Give more distance between groups
co2_plot_mixed + geom_jitter(width = 0.2)

Exercise: Two variable plots

Time: 8 min

# use dataset with two continuous variables
ggplot(___, aes(x = ___, y = ___)) +
    # finish the geom to create either a point, line, hex, smooth, or abline layer

# use dataset with one continuous and one discrete variable
ggplot(___, aes(x = ___, y = ___)) +
    # finish the geom to create either a boxplot, jitter, or col layer

Using a third (or more) variable

You can also add an additional dimension to the data by using other elements (colours, size, transparency, etc) of the graph to represent another variable. This is NOT the same thing as using 3-dimensionl (aka x, y, z axis) plots, which should be avoided unless absolutely necessary! Using colours to represent discrete groups is useful, or for using shading to represent a range in continuous values.

co2_plot_colour <- ggplot(CO2, aes(x = conc, y = uptake, colour = Treatment)) 

# Scatter plot
co2_plot_colour + geom_point()

# Line plot
co2_plot_colour + geom_line()

# Smoothing
co2_plot_colour + geom_smooth()

Or add a fourth variable.

# Scatter plot
co2_plot_colour + geom_point(aes(shape = Type))

# Line plot
co2_plot_colour + geom_line(aes(linetype = Type))

# Another line plot
co2_plot_colour + geom_path(aes(linetype = Type))

# Smoothing plot
co2_plot_colour + geom_smooth(aes(linetype = Type))

And it's easy to add another geoms!

# Three layers
co2_plot_colour + 
    geom_point(aes(shape = Type)) +
    geom_line(aes(linetype = Plant)) +
    geom_smooth(aes(size = Type))

Exercise: Three variable plots

Time: 8 min

# use dataset with either:
# - two continuous variables and one discrete
# - three continuous variables
# for last argument, choose either size, colour, alpha
ggplot(___, aes(x = ___, y = ___, ___ = ___)) +
    # finish the geom to create either a point or line layer

Axis titles and the theme

Let's get to making the plot prettier. There are many many options to customise the plot using the theme().

co2_plot_prettying <-
    ggplot(CO2, aes(
        x = conc,
        y = uptake,
        colour = paste(Treatment, Type)
    ) +
    geom_point() +

# Some pre-defined themes
co2_plot_prettying + theme_bw()
co2_plot_prettying + theme_minimal()
pretty_plot <- co2_plot_prettying +
    theme_classic() +
    scale_color_brewer(name = "Treatment and origin", palette = "Dark2") +
    # Find this information in ?CO2
    labs(x = "CO2 concentration (mL/L)",
         y = "CO2 update rate (umol/m2)") +
        # all axis lines, must use element_line
        axis.line = element_line(colour = "grey50", size = 0.5),
        # all axis text, must use element_text
        axis.text = element_text(family = "sans"),
        # all axis tick marks, use element_blank to remove
        axis.ticks = element_blank()

Exercise: Change theme items of the plot

Time: 10 min

# use dataset with two continuous variables
ggplot(___, aes(x = ___, y = ___)) +
    # finish the geom to create either a point, smooth, or line layer
    ___ +
    # choose either a minimal, dark, light, or classic defined theme
    ___ +
        # choose colours such as red, blue, black, grey, yellow, green
        # choose size from 2 to 8
        panel.grid.major = element_line(colour = ___, size = ___),
        # choose family such as sans, serif, Arial, Times New Romans
        axis.text = element_text(colour = ___, size = ___, familyl = ___)

Saving the plot

Now, if you want to save the plot, you can do that pretty easily!

ggsave("plant_co2_uptake.pdf", pretty_plot, width = 7, height = 5)

Exercise: Putting it all together

Time: Until end of session

  1. Create a ggplot, choosing three variables for the aes(), one for:
    • the x-axis
    • the y-axis
    • either size, colour, alpha, stroke, or fill
  2. Create two geom_ layers. The geom you use will depend on the variables and the specific aes() you choose above.
  3. Properly label the x and y axis with labs().
  4. Choose a pre-defined theme (theme_) and make two changes to it using theme().
  5. Save the plot with ggsave().

au-oc/content documentation built on May 21, 2019, 4:05 a.m.