ggplot2

Graphics with ggplot2

The package ggplot2 is used to produce graphics using a different system than base R plotting.

# install.packages(ggplot2)
library(ggplot2)

All such plots are composed of data and a mapping.

• Data: information you want to visualize
• Mapping: description of how the variables are mapped to aesthetic attributes.

There are five mapping components:

• Layer: a collect of geometric elements and statistical transformations.
  • Geometric elements are called geoms, they are what you see (points, lines, etc.)
  • Statistical transformations are called stats, they summarize the data (counting observations, making models)
• Scales: a map of the values in the data space to the aesthetic space.
  • Modifications of the color, shape, size
  • How to draw legends and axes
• Coordinate System: called coord, describe how data coordinates are mapped to the plane of the graphic.
  • Usually use Cartesian coordinate system.
• Facet: How to break up and display subsets of data as small multiples.
• Theme: control other points of display, like font size, and background color.

Getting started

Every ggplot has three key components:

1. Data.
2. A set of aesthetic mappings between variables and visual properties.
3. At least one layer which describes how to render each observation, usually created with a geom function.

Example:

ggplot(mpg,                         # Data
       aes(x = displ, y = hwy)) +   # aesthetic mapping
  geom_point()                      # geom layer 

Adjusting Aesthetic Attributes

Adjusting by variable

We can adjust aesthetic attributes by variable using the aes() function and the mapping argument.

Adjust the color for each point by class:

ggplot(mpg, aes(x = displ, y = hwy, color = class)) + 
  geom_point()

Adjust the shape for each point by drv:

ggplot(mpg, aes(x = displ, y = hwy, shape = drv)) + 
  geom_point()

Adjust the size of each point by cyl:

ggplot(mpg, aes(x = displ, y = hwy, size = cyl)) + 
  geom_point()

Adjusting for the entire plot

The plot changes based on if the color is assigned outside or inside the aes() function.

This is because aes() is used to map the data to an aesthetic property of the map. Inside of aes() we let the aesthetic argument (color, shape, etc) equal a variable in the data set. The different values for that variable control what happens to that aesthetic.

If we assign aesthetic value outside of the aes() function, we assign a value for the entire data set.

Change aesthetic attributes according to some variable:

ggplot(mpg, aes(displ, hwy)) + 
  geom_point(aes(color = class))

Attempt to change all points to blue:

ggplot(mpg, aes(displ, hwy)) + 
  geom_point(aes(color = "blue"))

Notice that this did not work because color = "blue" is inside aes().

Change all points to blue:

ggplot(mpg, aes(displ, hwy)) + 
  geom_point(color = "blue")

Faceting

Faceting creates tables of graphics by splitting the data into subsets and displaying the same graph for each subset.

You can facet using facet_wrap() which is more flexible and automatically determines how to organize the data. Or, you can manually determine the facet structure using facet_grid().

Facet wrap: let ggplot decide how many rows and columns to make.

# break data up by class
ggplot(mpg, aes(displ, hwy)) + 
  geom_point() + 
  facet_wrap(~class)

Facet grid: number of rows and columsn by number of unique values for each variable.

# Row variable ~ Column variable. 
ggplot(mpg, aes(displ, hwy)) + 
  geom_point() + 
  facet_grid(drv~class)

More Examples

# Example 1
ggplot(diamonds)+
  geom_boxplot(mapping=aes(y = price, x = cut, color = cut))

# Example 2
ggplot(diamonds)+
  geom_point(mapping=aes(y = price, x = depth, size = carat, 
                         color = cut), pch = 1)+
  labs(title = "A Title is On This Plot")

# Example 3
ggplot(data = diamonds) +
  geom_bar(mapping = aes(x = cut, fill=color), 
           position = "fill")

# Example 4
ggplot(diamonds, aes(x = clarity, group = cut, fill = cut)) +
  geom_bar(position = "dodge")

Resources

• https://ggplot2-book.org/index.html
• https://www.stat.auckland.ac.nz/~ihaka/downloads/Interface98.pdf
• https://hbctraining.github.io/Intro-to-R-flipped/cheatsheets/data-visualization-2.1.pdf

On Your Own

The code below generates an initial infection and plot for the SIR model. Use ggplot2 to recreate the plot. (You may need to dig through those resources and do some googling.)

set.seed(0)
nr <- 50
nc <- 50
p0 <- 0.1
sir.init <- matrix(rbinom(nr*nc, 1, p0), nr, nc)
image(sir.init, col = c("white","red"))