Day 5: Data visualization with ggplot2

Media day (week of 2023-10-30)

What is EDA?

EDA stands for Exploratory Data Analysis
EDA is the step of “getting to know” your data
You have already been doing some EDA by sorting the data and understanding what is in each column

What is EDA?

Another very useful tool for EDA is data visualization
Today we will learn how to visualize data using ggplot2
- ggplot2 is included in the tidyverse set of packages
- There are other ways to do this such as using plot(). But ggplot2 has more consistent syntax.

Image by Allison Horst

What makes a good plot?

How about this one?

Problems with this plot

Colors don’t correspond to contents (meat is green?)
3D doesn’t have any meaning, only makes plot more complicated
Hard for humans to visually compare area

What makes a good plot?

Simple
Easy to understand
Conveys a message

What makes a good plot?

Pie chart

drawing

Bar graph

drawing

Bar graphs are better than pie charts for comparing values!

`ggplot2` approach to plotting

Grammar of Graphics: all plots share a common “grammar”
1. Input data
2. Aesthetic mapping: How to represent the data visually
- aesthetic = 美的, mapping = 相当する. どんな美的な要素に相当するのかの指定
1. Geometry: What shape to use for the plot
Recommended Reading: ggplot2: Elegant Graphics for Data Analysis

Aesthetic mappings

Pie chart

drawing

Bar graph

drawing

How is population represented (“mapped”) in each plot?
What is the shape (“geometry”) of each plot?

Pie chart

drawing

Bar graph

drawing

How is population mapped in each plot?
- Mapped to area in pie chart, length in bar graph
What is the geometry of each plot?
- Circle (“pie”) in pie chart, column (“bar”) in bar graph

Reopen your project

We will continue to use the gapminder-analysis project in the data-analysis-course folder on your Desktop
Also create a file to write today’s code. You could call it data-viz-practice.R

Load `tidyverse`

Let’s load the tidyverse package with library()
- Note that you don’t need to use quotation marks for the package name
- Also load the scales package, which is for making plot labels

library(tidyverse)
library(scales)

Load data

Use the read_csv() function to load a spreadsheet as a dataframe

gapminder <- read_csv("data_raw/gapminder.csv")

Gapminder

This is a dataset of economic statistics from various countries over time, from https://gapminder.org
The meaning of some columns is obvious (country, continent, year), but not others
- pop: Population
- lifeExp: Life expectancy (寿命)
- gdpPercap: GDP per capita (一人当たりの国内総生産)

First plot

ggplot(data = gapminder)

First plot

ggplot(
  data = gapminder,
  mapping = aes(
    x = gdpPercap,
    y = lifeExp))

First plot

ggplot(
  data = gapminder,
  mapping = aes(
    x = gdpPercap,
    y = lifeExp)) +
  geom_point()

What does this plot tell us?

Challenge

Modify the code to show how life expectancy has changed over time

Add color

In the last challenge, we saw a general trend, but there could be more detail within certain groups, like continent or country
Let’s use color to show the continent
What is color in ggplot2?
- Color is an aesthetic mapping

Original plot

ggplot(
  data = gapminder,
  mapping = aes(
    x = year,
    y = lifeExp)) +
  geom_point()

Add color

ggplot(
  data = gapminder,
  mapping = aes(
    x = year,
    y = lifeExp,
    color = continent)) +
  geom_point()

Add color

ggplot(
  data = gapminder,
  mapping = aes(
    x = year,
    y = lifeExp,
    color = continent)) +
  geom_point()

Basic recipe

All plots can be made using the same basic formula:

ggplot(data = INPUT-DATA, mapping = aes(MAPPING)) +
  GEOMETRY

Basic recipe

Since data always comes first and mapping second, we can omit those names and make our code a little simpler:

ggplot(INPUT-DATA, aes(MAPPING)) +
  GEOMETRY

ggplot(
  gapminder, # DATA
  aes( # MAPPING
    x = year,
    y = lifeExp,
    color = continent)) +
  geom_point() # GEOMETRY

Plotting lines

Let’s try representing the data with a different geometry (shape)
This time we will use lines (geom_line())

Plotting lines

ggplot(
  gapminder, # DATA
  aes( # MAPPING
    x = year,
    y = lifeExp,
    color = continent)) +
  geom_line() # GEOMETRY

That looks a little odd. What is the problem?

Plotting lines

Lines connect points along the x-axis
But we only want to connect points within each country
We need to add another aesthic mapping for that, the group

Plotting lines

ggplot(
  gapminder,
  aes(
    x = year,
    y = lifeExp,
    color = continent)) +
  geom_line()

Plotting lines

ggplot(
  gapminder,
  aes(
    x = year,
    y = lifeExp,
    color = continent,
    group = country)) +
  geom_line()

Plotting lines

ggplot(
  gapminder,
  aes(
    x = year,
    y = lifeExp,
    color = continent,
    group = country)) +
  geom_line()

Adding layers with `+`

We can add additional plot layers using the + sign
For example, lets add points on top of the lines

Adding layers with `+`

ggplot(
  gapminder,
  aes(
    x = year,
    y = lifeExp,
    color = continent,
    group = country)) +
  geom_line()

Adding layers with `+`

ggplot(
  gapminder,
  aes(
    x = year,
    y = lifeExp,
    color = continent,
    group = country)) +
  geom_line() +
  geom_point()

Adding layers with `+`

ggplot(
  gapminder,
  aes(
    x = year,
    y = lifeExp,
    color = continent,
    group = country)) +
  geom_line() +
  geom_point()

Challenge

The ggtitle() adds a title to a plot. Use ggtitle() as another layer to add a title to the last plot.

Summary: Adding layers with `+`

Each layer modifies the plot, so you can build it gradually
Other things layers can do:
- Set the plot theme
- Change the axes
- Add labels
- and more!

Modify an aesthetic

So far, we have been mapping aesthetics to variables in the data
But you can also simply assign the same value to a particular aesthetic (such as color)
Do this by setting its value outside of mapping = aes()

Modify an aesthetic

ggplot(
  gapminder,
  aes(
    x = year,
    y = lifeExp,
    color = continent,
    group = country)) +
  geom_line() +
  geom_point()

Modify an aesthetic

ggplot(
  gapminder,
  aes(
    x = year,
    y = lifeExp,
    color = continent,
    group = country)) +
  geom_line() +
  geom_point(color = "black")

Modify an aesthetic

ggplot(
  gapminder,
  aes(
    x = year,
    y = lifeExp,
    color = continent,
    group = country)) +
  geom_line() +
  geom_point(color = "black")

Transformations

Let’s go back to the life expectancy vs. GDP graph again:

Transformations

Do you notice anything that makes this graph hard to read?

Transformations

Nearly all the points are crowded to the left, with only a few outliers on the right

We can make this easier to read by using a log-transformation

Transformations

ggplot(
  gapminder,
  aes(
    x = gdpPercap,
    y = lifeExp)) +
  geom_point()

Transformations

ggplot(
  gapminder,
  aes(
    x = gdpPercap,
    y = lifeExp)) +
  geom_point() +
  scale_x_log10()

Transformations

ggplot(
  gapminder,
  aes(
    x = gdpPercap,
    y = lifeExp)) +
  geom_point() +
  scale_x_log10()

Transformations

ggplot(
  gapminder,
  aes(
    x = gdpPercap,
    y = lifeExp)) +
  geom_point() +
  scale_x_log10(
    labels = label_comma()
  )

Multi-panel plots

ggplot2 can make multiple plots at once using facetting
Each facet is a mini-plot of some portion of the dataset
- Define the facet group with vars()
Let’s try this for some countries in Asia

Multi-panel plots

First, subset the data just to some countries in Asia (for this example)

gapminder_asia <- filter(
  gapminder,
  country %in% c("Japan", "China", "Korea, Rep.", "Taiwan"))

Multi-panel plots

ggplot(
  gapminder_asia,
  aes(
    x = year,
    y = lifeExp)) +
  geom_line()

Multi-panel plots

ggplot(
  gapminder_asia,
  aes(
    x = year,
    y = lifeExp)) +
  geom_line() +
  facet_wrap(vars(country))

Multi-panel plots

ggplot(
  gapminder_asia,
  aes(
    x = year,
    y = lifeExp)) +
  geom_line() +
  facet_wrap(vars(country))

Adjusting labels

When you “publish” your plot (this could be including it in a report for a class), it needs to be clearly labeled
- Whoever is reading your report won’t know what “lifeExp” means

Adjusting labels

ggplot(
  gapminder_asia,
  aes(
    x = year,
    y = lifeExp)) +
  geom_line() +
  facet_wrap(vars(country))

Adjusting labels

ggplot(
  gapminder_asia,
  aes(
    x = year,
    y = lifeExp)) +
  geom_line() +
  facet_wrap(vars(country)) +
  labs(
    x = "Year",
    y = "Life Expectancy (years)",
    title = "Figure 1"
  )

Adjusting labels

ggplot(
  gapminder_asia,
  aes(
    x = year,
    y = lifeExp)) +
  geom_line() +
  facet_wrap(vars(country)) +
  labs(
    x = "Year",
    y = "Life Expectancy (years)",
    title = "Figure 1"
  )

Saving your plot

If you need to you can save your plot in R, and write it out as an image file
Use the ggsave() function

Saving your plot

gapminder_asia_plot <- ggplot(
  gapminder_asia,
  aes(
    x = year,
    y = lifeExp)) +
  geom_line() +
  facet_wrap(vars(country)) +
  labs(
    x = "Year",
    y = "Life Expectancy (years)",
    title = "Figure 1"
  )

Saving your plot

ggsave(plot = gapminder_asia_plot, file = "gapminder_asia.pdf")

You can also save .jpg, .png, etc.

Summary

Think about what you want your graph to show
Aesthetic mapping: How to represent the data visually
Geometry: What shape to use for the plot
Build your plot using layers
Basic recipe: ggplot(INPUT-DATA, aes(MAPPING)) + GEOMETRY

Day 5: Data visualization with ggplot2

What is EDA?

What is EDA?

What makes a good plot?

Problems with this plot

What makes a good plot?

What makes a good plot?

ggplot2 approach to plotting

Aesthetic mappings

Reopen your project

Load tidyverse

Load data

Gapminder

First plot

First plot

First plot

Challenge

Add color

Original plot

Add color

Add color

Basic recipe

Basic recipe

Plotting lines

Plotting lines

Plotting lines

Plotting lines

Plotting lines

Plotting lines

Adding layers with +

Adding layers with +

Adding layers with +

Adding layers with +

Challenge

Summary: Adding layers with +

Modify an aesthetic

Modify an aesthetic

Modify an aesthetic

Modify an aesthetic

Transformations

Transformations

Transformations

Transformations

Transformations

Transformations

Transformations

Multi-panel plots

Multi-panel plots

Multi-panel plots

Multi-panel plots

Multi-panel plots

Adjusting labels

Adjusting labels

Adjusting labels

Adjusting labels

Saving your plot

Saving your plot

Saving your plot

Summary

`ggplot2` approach to plotting

Load `tidyverse`

Adding layers with `+`

Adding layers with `+`

Adding layers with `+`

Adding layers with `+`

Summary: Adding layers with `+`