DataVis Cheat Sheet

Hello
If you’re interested in data visualization, then working through this cheat sheet is a good place to start.
You’ll find examples and code that you can practice at home. The data that I use is available to you on your
computer right now. If you want to see a comprehensive list of practice data sets on your computer, simple
type data() into R Studio. As you install packages, the list of datasets available to you will increase.
The examples that I provide will 1) walk you through the basics of using ggplot to create a data visualization,
2) help you understand which plots to use, given different combinations of data that you might want to look
at, and 3) provide some code for beautiful examples of data visualization that you might want to use in your
own work.
This sheet won’t cover everything. For a more comprehensive overview of data visualization using R, please
visit www.learnmore365.com

Load packages
If you haven’t ever installed these packages, then you need to do so. You only ever have to install a package
on your computer once using the function install.packages("package_name")
Here are all of the packages used to develop the graphics in this cheat sheet. I’ve also included a line of code
that sets the theme for all plots to “black and white”

library(tidyverse)
library(ggridges)
library(patchwork)
library(viridis)
library(hrbrthemes)
library(gapminder)
theme_set(theme_bw())

Using ggplot2
To create graphics using ggplot2, you need to understand “the grammar of graphics”. All plots have three
principle components:

1. data
2. mapping
3. geometry

Data

The data is simply the dataset that you are using (nothing complicated about that).

1
Mapping

Mapping refers to how each variable that you are going to use in your visualization relates to a particular
aesthetic. This could be color, shape, size, x-axis, y-axis and others.

Geometry

Geometry refers to the type of plot that will be used to represent the dataset. For example, you might want
a boxplot, a histogram, a scatter plot, etc.
In the example below, I use the gapminder dataset (which is available to you once you’ve installed the
gapminder package) to represent 5 variables at the same time. Four numeric variables (GPD per capita,
Life expectancy, year and population size) and one categorical variable (continent).
Each of the numeric variables is mapped against a specific aesthetic and the categorical variable is used to
disaggregate the data into facets.

gapminder %>%
filter(continent %in% c("Africa", "Europe")) %>%
filter(gdpPercap < 30000) %>%
ggplot(aes(x= gdpPercap,
y = lifeExp,
size = pop,
color = year)) +
geom_point() +
facet_wrap(~continent) +
labs(title = "Life expectancy explained by GDP per capita",
x = "GDP per capita",
y = "Life expectancy")

2
 Life expectancy explained by GDP per capita
Africa Europe

80

pop
5e+07
1e+08
Life expectancy

60

year
2000
1990
1980
40
1970
1960

0 10000 20000 30000 0 10000 20000 30000

GDP per capita

How to visualize different combinations of data types

Here we are going to look at the starwars dataset. I’m going to provide you with examples of data
visualization using different combinations of numeric and categorical variables. In this subset of the starwars
data, I’ve included name as a unique identifier for each row (we won’t be using the name variable in the
analysis). height and mass are two numeric variables. gender and hair_color are categorical variables.

starwars %>%
select(name, height, mass, gender, hair_color) %>%
head()

## # A tibble: 6 x 5
## name height mass gender hair_color
## <chr> <int> <dbl> <chr> <chr>
## 1 Luke Skywalker 172 77 masculine blond
## 2 C-3PO 167 75 masculine <NA>
## 3 R2-D2 96 32 masculine <NA>
## 4 Darth Vader 202 136 masculine none
## 5 Leia Organa 150 49 feminine brown
## 6 Owen Lars 178 120 masculine brown, grey

3
Single numeric variable
Let’s start with a single numeric variable (height). In this figure we’ve created a histogram, a density plot,
a boxplot and a violin plot with the data. Here is the code and the outputs.

p1 <- starwars %>%

ggplot(aes(x = height)) +
geom_histogram(binwidth = 20,
show.legend = F,
alpha = .5) +
labs(title = "Histogram",
x = "Height",
y = "Count")

p2 <- starwars %>%

ggplot(aes(x = height)) +
geom_density(aes(fill = "blue"),
show.legend = F,
alpha = .5) +
labs(title = "Density plot",
x = "Height",
y = "Probability")

p3 <- starwars %>%

ggplot(aes(x = height)) +
geom_boxplot(show.legend = F,
alpha = .5) +
labs(title = "Boxplot",
x = "Height")

p3a <- starwars %>%

ggplot(aes(x = height, y = 1)) +
geom_violin(aes(fill = "blue"),
show.legend = F,
alpha = .5)+
labs(title = "Violin plot",
x = "Height")

(p1 /p2 | p3 / p3a) +

plot_annotation(title = "Single numeric variable",
theme = theme(plot.title = element_text(size = 18,
colour = "blue"))) +
theme(text = element_text('mono'))

4
 Single numeric variable
Histogram Boxplot
0.4
30
0.2
Count

20
0.0
10
−0.2

0 −0.4
100 200 100 150 200 250
Height Height

Density plot Violin plot

0.020
1.25
Probability

0.015
1.00
0.010 y
0.005 0.75

0.000
100 150 200 250 100 150 200 250
Height Height

One or more categorical variables

In this graphic I’ve included a bar plot for a single categorical variable. If a second categorical variable is
added we can disaggregate each bar by either stacking, grouping or presenting a proportion using the second
category.

p4 <- starwars %>%

drop_na(eye_color) %>%
filter(eye_color %in% c("black", "brown", "blue", "yellow")) %>%
ggplot(aes(x = eye_color)) +
geom_bar(stat = "count", alpha = 0.5) +
labs(title = "Barplot",
x = "Eye colour",
y = "Count")

p5 <- starwars %>%

drop_na(eye_color, gender) %>%
filter(eye_color %in% c("black", "brown", "blue", "yellow")) %>%
ggplot(aes(eye_color, fill = gender)) +
geom_bar(stat = "count", alpha = .5,
show.legend = F) +
labs(title = "Stacked barplot",
x = "Eye colour",
y = "Count")

5
p5a <- starwars %>%
drop_na(eye_color, gender) %>%
filter(eye_color %in% c("black", "brown", "blue", "yellow")) %>%
ggplot(aes(eye_color, fill = gender)) +
geom_bar(stat = "count", alpha = .5,
position="dodge",
show.legend = F)+
labs(title = "Grouped barplot",
x = "Eye colour",
y = "Count")

p5b <- starwars %>%

drop_na(eye_color, gender) %>%
filter(eye_color %in% c("black", "brown", "blue", "yellow")) %>%
ggplot(aes(eye_color, fill = gender)) +
geom_bar(stat = "count", alpha = .5,
position="fill",
show.legend = T) +
labs(title = "Percentage barplot",
x = "Eye colour",
y = "Count") +
theme(legend.position = "bottom")

((p4 | p5)/( p5a | p5b)) +

plot_annotation(title = "One or more categorical variable",
theme = theme(plot.title = element_text(size = 18,
colour = "blue"))) +
theme(text = element_text('mono'))

6
 One or more categorical variable
Barplot Stacked barplot
20 20

15 15
Count

Count
10 10
5 5
0 0
black blue brown yellow black blue brown yellow
Eye colour Eye colour

Grouped barplot Percentage barplot

15 1.00
0.75
Count

Count
10
0.50
5
0.25
0 0.00
black blue brown yellow black blue brown yellow
Eye colour Eye colour

gender feminine masculine

One numeric and two categorical variables

Here a single numeric variable can be represented by with a density plot or boxplot and then disaggregated
by one categorical variable using color or two categorical variables using color and facets.

p14a <- starwars %>%

drop_na(gender) %>%
ggplot(aes(height, fill = gender)) +
geom_boxplot(alpha = 0.3) +
labs(title = "Boxplot of a numeric variable",
subtitle = "disagregated by one categorical variable",
x = "Height") +
theme(legend.position = "none")

p14b <- starwars %>%

drop_na(gender) %>%
ggplot(aes(height, fill = gender)) +
geom_density(alpha = 0.3) +
labs(title = "Density plot of a numeric variable",
subtitle = "disagregated by one categorical variable",
x = "Height",
y = "Probability") +
theme(legend.position = "none")

7
p15 <- starwars %>%
drop_na(hair_color, gender) %>%
filter(hair_color %in% c("black", "brown")) %>%
ggplot(aes(height, fill = gender)) +
geom_density(alpha = 0.3) +
facet_wrap(~hair_color) +
labs(title = "Density plot of a numeric variable",
subtitle = "disagregated by two categorical variables",
x = "Height",
y = "Probability") +
theme(legend.position = "none")

p16 <- starwars %>%

filter(height > 140 & height < 200) %>%
drop_na(hair_color, gender) %>%
filter(hair_color %in% c("black", "brown")) %>%
ggplot(aes(height, fill = gender)) +
geom_boxplot(alpha = 0.3) +
facet_wrap(~hair_color) +
labs(title = "Boxplot of a numeric variable",
subtitle = "disagregated by two categorical variable",
x = "Height")+
theme(legend.position = "bottom")

((p14b/p14a)|(p15 / p16)) +
plot_annotation(title = "One numberic and two categorical variable",
theme = theme(plot.title = element_text(size = 18,
colour = "blue"))) +
theme(text = element_text('mono'))

8
 One numberic and two categorical variable
Density plot of a numeric variable Density plot of a numeric variable
disagregated by one categorical variable disagregated by two categorical variables
black brown

Probability
Probability

0.04 0.100
0.075
0.03 0.050
0.02 0.025
0.01 0.000
0.00 120 160 200 240 120 160 200 240
100 150 200 250 Height
Height
Boxplot of a numeric variable
Boxplot of a numeric variable
disagregated by two categorical variable
disagregated by one categorical variable
black brown
0.2
0.2
0.0 0.0
−0.2 −0.2
100 150 200 250 150 160 170 180 190 150 160 170 180 190
Height Height

gender feminine masculine

Two numeric and one categorical variable

Here we have a scatter plot of two numeric variables on the left. I’ve included a smoothed linear model with
standard error margins.
On the right we have the same data disaggregated by color and then by color and facets.

p6 <- starwars %>%

filter( mass < 250) %>%
ggplot(aes(x = height,
y = mass)) +
geom_point(size = 2,
alpha = 0.7) +
geom_smooth()+
labs(title = "Scatter plot",
subtitle = "with smoothed linear model",
x = "Height",
y = "Mass")

p7 <- starwars %>%

filter( mass < 250) %>%
drop_na(gender) %>%
ggplot(aes(height, mass, colour = gender)) +
geom_point(size = 2, show.legend = T) +
labs(title = "Scatter plot",

9
 subtitle = "disagregated by colour",
x = "Height",
y = "Mass")

p7a <- starwars %>%

filter( mass < 250) %>%
drop_na(gender) %>%
ggplot(aes(height, mass, colour = gender))+
geom_point(size = 2, show.legend = F)+
facet_wrap(~gender) +
labs(title = "Scatter plot",
subtitle = "disagregated by colour and facets",
x = "Height",
y = "Mass")

(p6 | p7 / p7a) +
plot_annotation(title = "Two numberic and one categorical variable",
theme = theme(plot.title = element_text(size = 18,
colour = "blue"))) +
theme(text = element_text('mono'))

## ‘geom_smooth()‘ using method = ’loess’ and formula ’y ~ x’

Two numberic and one categorical variable

Scatter plot Scatter plot
with smoothed linear model disagregated by colour
160
gender
150 120
Mass

80 feminine

40 masculine

100 100 150 200

Height
Mass

Scatter plot
disagregated by colour and facets
50
feminine masculine
160
120
Mass

0 80
40

100 150 200 100 150 200 100 150 200

Height Height

10
Lolipop graphic
Instead of plotting a simple boxplot for each of the categories, I’ve plotted individual data points, the average
for each category and then joined the average with a line that represents the average for the entire dataset.

chickwts %>%
group_by(feed) %>%
mutate(mean_by_feed = mean(weight)) %>%
ungroup() %>%
mutate(feed = fct_reorder(feed, mean_by_feed)) %>%
ggplot(aes(feed, weight, colour = feed,
show.legend = F)) +
coord_flip() +
geom_jitter(show.legend = F,
size = 4,
alpha = 0.2,
width = 0.05) +
stat_summary(fun = mean, geom = "point", size = 8, show.legend = F) +
geom_hline(aes(yintercept = mean(weight)),
colour = "gray70",
size = 0.9) +
geom_segment(aes(x = feed, xend = feed,
y = mean(weight), yend = mean_by_feed),
size = 2, show.legend = F) +
labs(title = "Weight of chickens by feed group",
x = "Feed",
y = "Weight of chickens") +
theme(legend.position = "none") +
theme_bw()

11
 Weight of chickens by feed group

sunflower

casein

meatmeal
Feed

soybean

linseed

horsebean

100 200 300 400

Weight of chickens

Using ridges
In this visualization I’ve shown distribution of temperatures that occurred within a given month (as a density
plot) and then compared each month by plotting the density plots of each month on the same canvas.

ggplot(lincoln_weather, aes(x = `Mean Temperature [F]`, y = `Month`, fill = ..x..)) +

geom_density_ridges_gradient(scale = 3, rel_min_height = 0.01,
alpha = 5) +
scale_fill_viridis(name = "Temp. [F]", option = "C") +
labs(title = 'Temperatures in Lincoln NE in 2016') +
theme_bw() +
theme(
legend.position="none",
panel.spacing = unit(0.1, "lines"),
strip.text.x = element_text(size = 8)
)

## Picking joint bandwidth of 3.37

12
 Temperatures in Lincoln NE in 2016

January

February

March

April

May
Month

June

July

August

September

October

November

December

0 25 50 75 100
Mean Temperature [F]

13