ggplot2

Author

Steen Flammild Harsted

Published

January 10, 2024

Getting Started

Create a new quarto document
Insert a code chunk and load 2 important libraries
Insert a new code chunk- Write source(here("scripts", "01_import.R")) in the chunk
Discuss what the code does and run it
Write a short headline to each code chunk

Code

# Load libraries
library(tidyverse)  # Data wrangling and plots
library(here)       # File control in project

# Load the cleaned soldiers dataset
source(here("scripts", "01_import.R"))

1 Guess the output

Look at and discuss the code below.

You’ll need to guess a little because you haven’t seen all the datasets and functions yet, but use your common sense! See if you can predict what each plot will look like before running the code.

ggplot(mpg, aes(cty, hwy)) + geom_point()

ggplot(diamonds, aes(carat, price)) + geom_point()

ggplot(economics, aes(date, unemploy)) + geom_line()

ggplot(mpg, aes(cty)) + geom_histogram()

2 Aesthetic mappings - aes()

Use the mpg dataset.

2.1 Explore the columntypes in the `mpg` dataset

Use one or more of: ?mpg, head(), glimpse(), summary(), and/or skim()

Code

?mpg
mpg %>%  head()
mpg %>%  glimpse()
mpg %>% skimr::skim()  # or library(skimr) and then mpg %>% skim()

2.2 Experiment with the `colour`, `shape`, and `size` aesthetics.

Use this basic code, and add/change the colour, shape, and size aesthetics.

ggplot(mpg, aes(cty, hwy)) + 
  geom_point()

What happens when you map continuous variables to one or more of the aesthetics?
What about categorical variables?
What happens when you use multiple aesthetics in a plot?

Code

## Examples

# Categorial
ggplot(mpg, aes(cty, hwy, colour = class)) + 
  geom_point()

# Continuous 
ggplot(mpg, aes(cty, hwy, size = displ)) + 
  geom_point()

# Continuous 
ggplot(mpg, aes(cty, hwy, color = hwy)) +  # Notice hwy is mapped to both y axis and color
  geom_point()

## A continuous variable doesn't work for shape
ggplot(mpg, aes(cty, hwy, shape = displ)) + 
  geom_point()

# Multiple Categorical - a legend for each aesthetic is created
ggplot(mpg, aes(cty, hwy, colour = class, shape = fl)) + 
  geom_point()

2.3 What’s gone wrong with this code? Why are the points not blue?

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

Code

# If you want all the points to be colored blute, then color = "blue" must be placed outside the aes() function.
ggplot(data = mpg) + 
  geom_point(mapping = aes(x = displ, y = hwy),
             color = "blue")

3 `geom_point()` and `geom_smooth()`

Use your `soldiers` dataset

Explore the relationship between heightcm and weightkg using geom_point()

Code

soldiers %>% 
  ggplot(aes(x = heightcm, y = weightkg))+
  geom_point()

Color the points according to BMI

Code

soldiers %>% 
  ggplot(aes(x = heightcm, y = weightkg, color = BMI))+
  geom_point()

Color the points according to category

Code

soldiers %>% 
  ggplot(aes(x = heightcm, y = weightkg, color = category))+
  geom_point()

Explore the relationship between heightcm and weightkg using geom_point() and geom_smooth()

Code

soldiers %>% 
  ggplot(aes(x = heightcm, y = weightkg))+
  geom_point()+
  geom_smooth(method = "lm")

color by sex

Code

soldiers %>% 
  ggplot(aes(x = heightcm, y = weightkg, color = sex))+
  geom_point()+
  geom_smooth(method = "lm")

4 `facet_wrap()`

What arguments can you use to control how many rows and columns appear in the output?
What does the scales argument in facet_wrap() do? When might you use it?

Explore the relationship between heightcm and weightkg

Use geom_point() and geom_smooth()
facet by sex
color the points by category

Code

soldiers %>% 
  ggplot(aes(x = heightcm, y = weightkg))+
  geom_point(aes(color = category))+
  geom_smooth(method = "lm")+
  facet_wrap(~sex)

5 `geom_bar()`

Use `geom_bar()` to explore how many soldiers of each `race` there is

Code

soldiers %>% 
  ggplot(aes(x = race))+
  geom_bar()

Use `geom_bar()` to explore how many soldiers are at each `Installation`

Code

soldiers %>% 
  ggplot(aes(x = Installation))+
  geom_bar()

# OR - Whats the difference?

soldiers %>% 
  ggplot(aes(y = Installation))+
  geom_bar()

Use the fill aestetic to color the Installation bars according to race

Code

soldiers %>% 
  ggplot(aes(y = Installation, fill = race))+
  geom_bar()

Change something so that you can visually evaluate if race is equally distributed across the Installations

Code

soldiers %>% 
  ggplot(aes(y = Installation, fill = race))+
  geom_bar(position = "fill",   #  All bars have full length/height - this makes it easier to see proportional differences between groups
           color = "black"      #  Adds a black line around each box.
           )

6 `geom_boxplot()`

Use `geom_boxplot` to explore `weightkg` across the different Installations

Code

soldiers %>% 
  ggplot(aes(x = Installation, y = weightkg))+
  geom_boxplot()

# OR

soldiers %>% 
  ggplot(aes(y = Installation, x = weightkg))+
  geom_boxplot()

Remove the outliers from the boxplot (read the documentation)
Add some jittered points to give an impression of the nr of soldiers at each installation
Give the jittered points some transparency (decrease alpha) to avoid overplotting

Code

soldiers %>% 
  ggplot(aes(y = Installation, x = weightkg))+
  geom_boxplot(
    outlier.shape = NA)+
  geom_jitter(
    height = 0.2,
    alpha = 0.1)

* Use facet_wrap() to split on sex have the plots placed in one column

Code

soldiers %>% 
  ggplot(aes(y = Installation, x = weightkg))+
  geom_boxplot(
    outlier.shape = NA)+
  geom_jitter(
    height = 0.2,
    alpha = 0.1)+
  facet_wrap(~sex, ncol = 1)

7 Annotations (title, legends, etc.)

Read the documentation for `labs()`

What can we control with this function?

Use `labs()` and recreate this plot

To create a completely blank plot just type ggplot()

Code

ggplot()+
  labs(x = "X-axis",
       y = "Y-axis",
       title = "Title",
       subtitle = "Subtitle",
       caption = "Caption")

8 Themes

What theme was used for this plot?

To recreate the basic plot use: ggplot(mpg, aes(hwy, cty))+ geom_point()
Try another theme. Type theme_ and try some of the suggestions.

Want more?

Use install.packages() to download the ggthemes package. When you have done that add library(ggthemes) to the code chunk where you call your libraries and execute the line.
Watch the gallery

9 Explore `soldiers` further

Together with your neighbor
Come up with a research question for the dataset
Discuss what variables to map
Discuss what geoms to use
Fix the labels in your plot (x, y, title, etc..)
When you are done - show your plot to another group and take a look at their plot
Try to work out the code they must have used to create that plot

10 Saving plots

In your project folder, create a new folder for saving plots and/or tables (e.g. “plots”)

read the documentation for `ggsave()`

What does the arguments I have specified below do?
Are they different from the defaults?

ggsave(filename = here("[YOUR FOLDER]", "[THE NAME OF YOUR FILE].png"), # .png .pdf .jpg are typical options
       plot = [The name of the ggplot object],                      # 
       dpi = "retina", 
       device = NULL                                                # Why can we leave this as NULL?
       )

Create a simple plot and save it to your folder using `ggsave()`

ggplot(mpg, aes(hwy, cty))+
  geom_point()+
  theme_classic()

ggsave(filename = here("plots", "my_plot.png"),
       plot = last_plot(),
       dpi = "retina")

11 color and fill scales

Put a different fill scale on this plot

starwars %>% 
  ggplot(aes(y = skin_color, fill = sex))+
  geom_bar()

Code

starwars %>% 
  ggplot(aes(y = skin_color, fill = sex))+
  geom_bar() +
  scale_fill_brewer(type = "qual", palette = 3)

Put a different color scale on this plot

mtcars %>%
  ggplot(aes(x = mpg, y = disp, color = hp)) +
  geom_point(size = 6)

Code

mtcars %>%
  ggplot(aes(x = mpg, y = disp, color = hp)) +
  geom_point(size = 6) +
  scale_color_continuous(type = "viridis", option = "magma")

Using `diamonds`, recreate the R code necessary to generate the following graphs.

Using `diamonds`, recreate the R code necessary to generate the following graph.

Using `diamonds`, recreate the R code necessary to generate the following graphs.

HINT

HINT: Think position=?????

12 Overplotting

Fix this plot

change alpha
change shape

diamonds %>% 
  filter(x>3 & x<9) %>% 
  ggplot(aes(x = x, y = price))+
  geom_point()

Code

# alpha and point
diamonds %>% 
  filter(x>3 & x<9) %>% 
  ggplot(aes(x = x, y = price))+
  geom_point(alpha = 0.2,
             shape = ".") # This shape gives each point the size of a pixel

13 Arranging Plots

Install the `patchwork` package and load it

Use install.packages() to download the patchwork package. When you have done that add library(patchwork) to the code chunk where you call your libraries and execute the line.

Run this code and then recreate the plot below

p <- ggplot(diamonds)
pf <- ggplot(diamonds %>% filter(carat<3))
p1 <- p + geom_bar(aes(x = cut, fill = clarity), position = "fill") +labs(title = "p1")
p2 <- p + geom_bar(aes(x = cut, fill = clarity)) +labs(title = "p2")
p3 <- pf + geom_histogram(aes(x = carat, fill = clarity), binwidth = 0.1, position = "fill", na.rm = TRUE) +labs(title = "p3")
p4 <- pf + geom_histogram(aes(x = carat, fill = clarity), binwidth = 0.1, position = "dodge", na.rm = TRUE) +labs(title = "p4")

Code

(p1|p2/(p3+p4)) + plot_layout(guide = "collect")

Read the documentation for `plot_annotation()`

What does the function overall do?
What does the argument tag_levels = 'A' do?

Tag the plots with roman numerals, and suffix the numerals with a “.”

Code

(p1|p2/(p3+p4)) + 
  plot_layout(guide = "collect") +
  plot_annotation(tag_levels = "I",
                  tag_suffix = ".")

Add a catchy title

Code

(p1|p2/(p3+p4)) +
  plot_layout(guide = "collect") +
  plot_annotation(tag_levels = "I",
                  tag_suffix = ".", 
                  title = "A catchy title")

Remove the small plot titles (p1, p2, p3, p4) from the four plots

You need to use the & symbol to do this the smart way You need to use the labs() function You need to set an argument in the labs() function to NULL

Code

(p1|p2/(p3+p4)) +
  plot_layout(guide = "collect") +
  plot_annotation(tag_levels = "I",
                  tag_suffix = ".", 
                  title = "A catchy title") &
  labs(title = NULL)

Apply the `theme_void()` and the viridis magma fill scale

Code

(p1|p2/(p3+p4)) +
  plot_layout(guide = "collect") +
  plot_annotation(tag_levels = "I",
                  tag_suffix = ".",
                  title = "A catchy title") & 
  theme_void() &
  scale_fill_viridis_d(option = "magma")

Save the plot to your plot folder

Code

ggsave(filename = here("plots", "my_patchwork_plot.png"),
       plot = last_plot(),
       dpi = "retina")

Getting Started

1 Guess the output

Look at and discuss the code below.

2 Aesthetic mappings - aes()

2.1 Explore the columntypes in the mpg dataset

2.2 Experiment with the colour, shape, and size aesthetics.

2.3 What’s gone wrong with this code? Why are the points not blue?

3 geom_point() and geom_smooth()

Use your soldiers dataset

4 facet_wrap()

5 geom_bar()

Use geom_bar() to explore how many soldiers of each race there is

Use geom_bar() to explore how many soldiers are at each Installation

6 geom_boxplot()

Use geom_boxplot to explore weightkg across the different Installations

7 Annotations (title, legends, etc.)

Read the documentation for labs()

Use labs() and recreate this plot

8 Themes

What theme was used for this plot?

Want more?

9 Explore soldiers further

10 Saving plots

read the documentation for ggsave()

Create a simple plot and save it to your folder using ggsave()

11 color and fill scales

Put a different fill scale on this plot

Put a different color scale on this plot

Using diamonds, recreate the R code necessary to generate the following graphs.

Using diamonds, recreate the R code necessary to generate the following graph.

Using diamonds, recreate the R code necessary to generate the following graphs.

12 Overplotting

Fix this plot

13 Arranging Plots

Install the patchwork package and load it

Run this code and then recreate the plot below

Read the documentation for plot_annotation()

Tag the plots with roman numerals, and suffix the numerals with a “.”

Add a catchy title

Remove the small plot titles (p1, p2, p3, p4) from the four plots

Apply the theme_void() and the viridis magma fill scale

Save the plot to your plot folder

2.1 Explore the columntypes in the `mpg` dataset

2.2 Experiment with the `colour`, `shape`, and `size` aesthetics.

3 `geom_point()` and `geom_smooth()`

Use your `soldiers` dataset

4 `facet_wrap()`

5 `geom_bar()`

Use `geom_bar()` to explore how many soldiers of each `race` there is

Use `geom_bar()` to explore how many soldiers are at each `Installation`

6 `geom_boxplot()`

Use `geom_boxplot` to explore `weightkg` across the different Installations

Read the documentation for `labs()`

Use `labs()` and recreate this plot

9 Explore `soldiers` further

read the documentation for `ggsave()`

Create a simple plot and save it to your folder using `ggsave()`

Using `diamonds`, recreate the R code necessary to generate the following graphs.

Using `diamonds`, recreate the R code necessary to generate the following graph.

Using `diamonds`, recreate the R code necessary to generate the following graphs.

Install the `patchwork` package and load it

Read the documentation for `plot_annotation()`

Apply the `theme_void()` and the viridis magma fill scale