TOOL

Changing Scales With ggplot()

When creating a data visualization, you'll presumably want to control aspects like which colors
are used in the plot and what tick marks appear on the x- and y-axes. You can do so using scale
functions, which allow you to adjust the scales, or values, used for colors, shapes, sizes, numbers,
and more. The ggplot2 package offers a variety of scale functions; we summarize some of the most
commonly used scale functions here.

For each of the following examples, we've used data from the storms.csv file you used during the
course. To set that data up for use with this tool, use the following code:

library(tidyverse)
# Read in the storm data
storms <- read.csv("storms.csv")
# Set the storm category to be a factor
storms$Category <- factor(storms$Category, levels = -1:5)
# Set the measurement date/time to be a factor
storms$Date <- factor(storms$Date, levels = unique(storms$Date))
# Create data set with observations only for Hurricanes Katrina, Sandy,
# and Wilma
sampleStorms <- storms %>% filter(Name %in% c("Katrina", "Sandy", "Wilma"))

Changing Color, Fill, and Shape for Categorical Variables

When you create an aesthetic mapping that distinguishes categories, R will apply default colors
or shapes, but you may want to change them. You can use scale_color_manual() to change
the colors of categories within a scatterplot, scale_fill_manual() to change the colors of
categories within a bar chart, and scale_shape_manual() to change the shape of categories
within a scatterplot. The characteristics you can manually change via arguments to these functions
are value, labels, and name. Together, these arguments allow you to specify the color or shape
associated with each category, change the legend title, and change the legend labels. You can
specify colors via their hex codes, which are six-digit codes used to represent colors based on
their red, blue, and green components, or via a string (e.g., "red," "black," "blue"). You can specify
shapes by their “point characteristic,” or pch, which you can look up in the help menu or online.

Data Cleaning With the Tidyverse © 2021 Cornell University 1

Cornell Bowers College of Computing and Information Science
ggplot(data = sampleStorms, aes(x = TS_
diameter, y = Pressure, color = Name)) +
geom_point() +
scale_color_manual(values = c('Katrina' =
'#393f47', 'Sandy' = '#b31b1b', 'Wilma'
= '#fbb040'),
labels = c('Katrina' = "KATRINA", 'Sandy'
= "SANDY", 'Wilma' = "WILMA"),
name = "Storm Name")

ggplot(sampleStorms, aes(x = Name, fill =

Category)) +
geom_bar() +
scale_fill_manual(values = c('-1' =
'#b31b1b', '0' = '#cecece', '1' =
'#393f47', '2' = '#92b2c4', '3' =
'#fbb040'),
labels = c('-1' = 'tropical depression',
'0' = 'tropical storm', '1' = 'category
1 hurricane', '2' = 'category 2
hurricane', '3' = 'category 3
hurricane'),
name = 'Storm Category')

ggplot(data = sampleStorms, aes(x = TS_

diameter, y = Pressure, shape = Category)) +
geom_point() +
scale_shape_manual(values = c('-1' = 15,
'0' = 16, '1' = 17, '2' = 18, '3' = 25),
labels = c('-1' = 'tropical depression',
'0' = 'tropical storm', '1' = 'category 1
hurricane','2' = 'category 2 hurricane',
'3' = 'category 3 hurricane'),
name = 'Storm Category')

Data Cleaning With the Tidyverse © 2021 Cornell University 2

Cornell Bowers College of Computing and Information Science
Changing Size and Transparency for Quantitative Variables
When you want to distinguish characteristics within a gradient, you may want R to scale a particular
characteristic based on the value of the variable. When these values exist along a spectrum, you
can use scale_size() to scale the point sizes, scale_alpha() to scale the point transparencies,
and scale_color_gradient() to scale the point colors. These functions use the arguments names
and labels to adjust the legend, just as the previous functions did. These functions, however, also
take the argument breaks to identify which sizes, transparencies, or colors should be marked in
the legend. Additionally, scale_size() and scale_alpha() take the argument range to specify
the high and low ends of the scales, while scale_color_gradient() takes the arguments low and
high to do the same.

ggplot(data = sampleStorms, aes(x = TS_

diameter, y = Pressure, size = Wind)) +
geom_point() +
scale_size(breaks = c(40, 60, 80, 100),
labels = c("40 mph", "60 mph", "80 mph",
"100 mph"),
name = "Wind Speed",
range = c(1,6))

ggplot(data = sampleStorms, aes(x = TS_

diameter, y = Pressure, alpha = Wind)) +
geom_point() +
scale_alpha(breaks = c(40, 60, 80, 100),
labels = c("40 mph", "60 mph", "80 mph",
"100 mph"),
name = "Wind Speed",
range = c(0.4,0.8))

ggplot(data = sampleStorms, aes(x = TS_

diameter, y = Pressure, color = Wind)) +
geom_point() +
scale_color_gradient(low = "blue",
high = "red",
breaks = c(40, 60, 80, 100),
labels = c("40 mph", "60 mph", "80 mph",
"100 mph"),
name = "Wind Speed")

Data Cleaning With the Tidyverse © 2021 Cornell University 3

Cornell Bowers College of Computing and Information Science
Changing Axis Scales
R creates scales for the x- and y-axes based on the range of values of the corresponding variables.
For example, the default scale for TS_diameter runs from 0 to 1000 in increments of 250. You might
want to adjust these scales. For example, if the variable on the x-axis is continuous, you can adjust
the x-axis scale with the function scale_x_continuous(). Below, the TS_diameter scale is adjusted
so that it runs from 0 to 1000 but in increments of 200.

ggplot(data = sampleStorms, aes(x = TS_

diameter, y = Pressure)) +
geom_point() +
scale_x_continuous(breaks = c(0, 200,
400, 600, 800, 1000),
labels = c("0", "200", "400", "600",
"800", "1000"))

Setting Axis and Plot Titles

You can use the labs() function to set the axis names, the plot title, and even a subtitle. You can
also adjust the name(s) of any legend(s) you created. In the example below, the points are colored
according to storm (color = "Name"). Inside labs(), we can set color = "Storm Name" so that the
legend is titled "Storm Name."

Relationship Between Storm Size and Air Pressure

ggplot(data = sampleStorms, aes(x = TS_ For Three Major Storms (Katrina, Sandy, Wilma)

diameter, y = Pressure, color = Name)) +

1000
geom_point() +
labs(x = "Tropical Storm Diameter",
980
y = "Air Pressure", Storm Name
Air Pressure

Katrina
color = "Storm Name", Sandy
960
title = "Relationship Between Storm Wilma

Size and Air Pressure",

940
subtitle = "For Three Major Storms
(Katrina, Sandy, Wilma)")
920
0 250 500 750 1000
Tropical Storm Diameter

Data Cleaning With the Tidyverse © 2021 Cornell University 4

Cornell Bowers College of Computing and Information Science
Combining Scales
You can use any combination of the functions above to produce visualizations that are more advanced.

ggplot(data = sampleStorms, aes(x = TS_

diameter, y = Pressure, color = Wind,
shape = Name)) +
geom_point() +
scale_color_gradient(low = "blue", high =
"red",
breaks = c(40, 60, 80, 100),
labels = c("40 mph", "60 mph", "80 mph",
"100 mph")) +
scale_shape_manual(values = c('Katrina' =
15, 'Sandy' = 16, 'Wilma' = 17)) +
labs(x = "Tropical Storm Diameter",
y = "Air Pressure",
color = "Wind Spped",
shape = "Storm Name",
title = "Relationship Between Storm
Size, Air Pressure, and Wind Speed",
subtitle = "For Three Major Storms
(Katrina, Sandy, Wilma)")

Data Cleaning With the Tidyverse © 2021 Cornell University 5

Cornell Bowers College of Computing and Information Science