5A: Data Visualization with ggplot2

Readings

From R for Data Science by Drs. Hadley Wickham, Garrett Grolemund, and Mine Çetinkaya-Rundel:

From ggplot2: Elegant Graphics for Data Analysis (3e) by Dr. Hadley Wickham:

Also check out Data Tidying from Data Science with R by Dr. Garrett Grolemund for more in-depth explanation of tidy data.

Topics

The tidyverse
Tidy data
Tibble
ggplot2 and the grammar of graphics
Data visualization with ggplot2

The `tidyverse`

The tidyverse package is a collection of packages for data science developed by Hadley Wickham and his team at Posit:
- ggplot2 for data visualization
- dplyr for data manipulation
- tidyr for getting data in “tidy” format
- tibble for enhancing data frames
- readr for importing rectangular data (like csv, tsv, and fwf)
- haven for importing and exporting SPSS, Stata and SAS files
- purrr for functional programming
- stringr for working with strings
- forcats for handling categorical data
- lubridate for working with date-times
We can install tidyverse as follows

install.packages("tidyverse")

To use tidyverse, we need to load it as follows

library(tidyverse)

We may encounter some variation of following message and that is okay

# ── Attaching core tidyverse packages ─────────────────────── tidyverse 2.0.0 ──
# ✔ dplyr     1.1.4     ✔ readr     2.1.5
# ✔ forcats   1.0.0     ✔ stringr   1.5.1
# ✔ ggplot2   4.0.0     ✔ tibble    3.2.1
# ✔ lubridate 1.9.4     ✔ tidyr     1.3.1
# ✔ purrr     1.0.2     
# ── Conflicts ───────────────────────────────────────── tidyverse_conflicts() ──
# ✖ tidyr::expand() masks Matrix::expand()
# ✖ dplyr::filter() masks stats::filter()
# ✖ dplyr::lag()    masks stats::lag()
# ✖ tidyr::pack()   masks Matrix::pack()
# ✖ dplyr::select() masks MASS::select()
# ✖ tidyr::unpack() masks Matrix::unpack()
# ℹ Use the conflicted package to force all conflicts to become errors

Tidy data

Tabular data are often organized in a wide format or a long format.
The wide format typically uses more columns and requires less repetition.

On the other hand, the long format tends to be less compacted with more rows.

Tabular data is tidy if each variable forms a column, each observation forms a row, and each cell is a single measurement. Tidy data is often in a long format.

tidyverse believes that tidy data in a long format is ideal for doing data science with R, because it is easier to extract every value of a variable to build a plot or to compute a summary statistic.

💻 Hands-On

The heights in centimeters and weights in kilograms of Amy and Ben are measured at ages 11, 14, and 18. The data are organized in a long format and a wide format. If we want to convert heights to inches and weights to pounds, which format is easier to work with?

long_df <- data.frame(
  name   = c('Amy', 'Amy', 'Amy', 'Ben', 'Ben', 'Ben'),
  age    = c(11, 14, 18, 11, 14, 18),
  height = c(144, 160, 167, 146, 158, 165),
  weight = c(36, 50, 58, 38, 48,55)
)

wide_df <- data.frame(
  name   = c('Amy', 'Ben'),
  height_11 = c(144, 146),
  height_14 = c(160, 158),
  height_18 = c(167, 165),
  weight_11 = c(36, 38),
  weight_14 = c(50, 58),
  weight_18 = c(58, 55)
)

Answer

Extracting, transforming, and modifying variables can be easier with the long format compared to the wide format.

# With long format
long_df$height

# With wide format
wide_df[, c('height_11', 'height_14', 'height_18')]

Tibble

A tibble is an enhanced version of base R data frame developed in tidyverse.
One noticeable difference is that a tibble has better printing capabilities.
For a comprehensive list of Differences between tibbles vs data.frames, check out the video of Dr. Johsua French.

💻 Hands-On

Try the following code and describe the difference between a traditional data frame and a tibble.

# Loan data
loan <- read.csv(url('https://raw.githubusercontent.com/hungtong/DS-01-101/refs/heads/main/dataset/loan50.csv'))

# Traditional data frame
loan

# Tibble
as_tibble(loan)

Answer

Since loan has 50 observations and 18 variables, printing loan as a data frame will take a lot of space in the console. Tibble presents the data more compactly and descriptively.

# # A tibble: 50 × 18
#    state emp_length  term homeownership annual_income verified_income
#    <chr>      <int> <int> <chr>                 <int> <chr>          
#  1 NJ             3    60 rent                  59000 Not Verified   
#  2 CA            10    36 rent                  60000 Not Verified   
#  3 SC            NA    36 mortgage              75000 Verified       
#  4 CA             0    36 rent                  75000 Not Verified   
#  5 OH             4    60 mortgage             254000 Not Verified   
#  6 IN             6    36 mortgage              67000 Source Verified
#  7 NY             2    36 rent                  28800 Source Verified
#  8 MO            10    36 mortgage              80000 Not Verified   
#  9 FL             6    60 rent                  34000 Not Verified   
# 10 FL             3    60 mortgage              80000 Source Verified
# # ℹ 40 more rows
# # ℹ 12 more variables: debt_to_income <dbl>, total_credit_limit <int>,
# #   total_credit_utilized <int>, num_cc_carrying_balance <int>,
# #   loan_purpose <chr>, loan_amount <int>, grade <chr>, interest_rate <dbl>,
# #   public_record_bankrupt <int>, loan_status <chr>, has_second_income <lgl>,
# #   total_income <int>
# # ℹ Use `print(n = ...)` to see more rows

Fuel economy data `mpg`

For this lesson, we will work with the data set mpg from ggplot2.
mpg contains information about the fuel economy of popular car models between 1999 and 2008 collected by the Environmental Protection Agency (EPA). The data set is a subset from https://fueleconomy.gov/.

library(ggplot2)

mpg

# A tibble: 234 × 11
   manufacturer model      displ  year   cyl trans drv     cty   hwy fl    class
   <chr>        <chr>      <dbl> <int> <int> <chr> <chr> <int> <int> <chr> <chr>
 1 audi         a4           1.8  1999     4 auto… f        18    29 p     comp…
 2 audi         a4           1.8  1999     4 manu… f        21    29 p     comp…
 3 audi         a4           2    2008     4 manu… f        20    31 p     comp…
 4 audi         a4           2    2008     4 auto… f        21    30 p     comp…
 5 audi         a4           2.8  1999     6 auto… f        16    26 p     comp…
 6 audi         a4           2.8  1999     6 manu… f        18    26 p     comp…
 7 audi         a4           3.1  2008     6 auto… f        18    27 p     comp…
 8 audi         a4 quattro   1.8  1999     4 manu… 4        18    26 p     comp…
 9 audi         a4 quattro   1.8  1999     4 auto… 4        16    25 p     comp…
10 audi         a4 quattro   2    2008     4 manu… 4        20    28 p     comp…
# ℹ 224 more rows

The data set contains the following categorical variables:
- manufacturer name
- model name
- year of manufacture
- trans - type of transmission
- drv - type of drive train, front-wheel f, rear-wheel r, or four-wheel 4
- fl - fuel type
- class - type of car

The data set contains the following numerical variables:
- displ - engine displacement (size), in litres
- cyl - number of cylinders
- cty - miles per gallon for city driving
- hwy - miles per gallon for highway driving

For this lesson, we are interested in the relationship of engine size and fuel economy. Note that engine size can be indicated by displ.

💻 Hands-On

Try the following code and describe the difference between str() and glimpse(). Note that str() is from base R and glimpse() is from the tidyverse.

# Base R
str(mpg)

# Tidyverse
glimpse(mpg)

Answer

glimpse() is similar to str() but seems to be formatted better.

# Rows: 234
# Columns: 11
# $ manufacturer <chr> "audi", "audi", "audi", "audi", "audi", "audi", "audi", "aud…
# $ model        <chr> "a4", "a4", "a4", "a4", "a4", "a4", "a4", "a4 quattro", "a4 …
# $ displ        <dbl> 1.8, 1.8, 2.0, 2.0, 2.8, 2.8, 3.1, 1.8, 1.8, 2.0, 2.0, 2.8, …
# $ year         <int> 1999, 1999, 2008, 2008, 1999, 1999, 2008, 1999, 1999, 2008, …
# $ cyl          <int> 4, 4, 4, 4, 6, 6, 6, 4, 4, 4, 4, 6, 6, 6, 6, 6, 6, 8, 8, 8, …
# $ trans        <chr> "auto(l5)", "manual(m5)", "manual(m6)", "auto(av)", "auto(l5…
# $ drv          <chr> "f", "f", "f", "f", "f", "f", "f", "4", "4", "4", "4", "4", …
# $ cty          <int> 18, 21, 20, 21, 16, 18, 18, 18, 16, 20, 19, 15, 17, 17, 15, …
# $ hwy          <int> 29, 29, 31, 30, 26, 26, 27, 26, 25, 28, 27, 25, 25, 25, 25, …
# $ fl           <chr> "p", "p", "p", "p", "p", "p", "p", "p", "p", "p", "p", "p", …
# $ class        <chr> "compact", "compact", "compact", "compact", "compact", "comp…

`ggplot2` and the grammar of graphic

ggplot2 is an R package for data visualization which implements the grammar of graphics introduced by Leland Wilkinson.
While base R graphics we have seen are generally fast, ggplot2 is versatile and widely used for making elegant and beautiful graphics.
ggplot2 is good for visualizing the relationship between multiple variables.
The grammar of graphic of Leland Wilkinson defines a statistical graph based on the following layers (components):

ggplot2 builds a plot by adding multiple layers, each of which has a specific role. At the most basic level, a ggplot2 consist of:
- The data set to visualize
- Aesthetics that map variables to visual properties
- Geometry to represent the observations
Common aesthetics include:
- x for the x-axis
- y for the y-axis
- fill for interior color (bars, areas, polygons, etc.)
- color for line or point color
- size for the size of the observation
- alpha for adjusting transparency
- shape for the shape of the observation
Common geometries include:
- geom_point for scatter plot
- geom_bar for bar plot
- geom_line for line plot
- geom_smooth for smoothing line
- geom_histogram for histogram
- geom_density for density curve
- geom_boxplot for box plot
- geom_violin for violin plot
Facets divide data into groups and display multiple plots to compare them.
Statistics provide summaries and transformations.
Coordinates control the position of data points: Cartesian, fixed-aspect ratio, or polar coordiate.
Theme controls the appearance of the plot and annotations.
The R Graph Gallery provides a collection of graphs made with base R and ggplot2.

Scatter plot `geom_point()`

The first layer of the plot is the data set to visualize.

ggplot(mpg)

The second layer is the mapping of aesthetics.

ggplot(mpg, mapping = aes(x = displ, y = hwy))

The third layer indicates which geometry to represent the observations.

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

Color, size, and transparency can be adjusted

ggplot(mpg, mapping = aes(x = displ, y = hwy)) + 
  geom_point(color = 'steelblue', size = 3, alpha = 0.7)

For scatter plots, adding a smoothing line with geom_smooth() can help us see the trend.

ggplot(mpg, mapping = aes(x = displ, y = hwy)) + 
  geom_point(color = 'steelblue') +
  geom_smooth()

`geom_smooth()` using method = 'loess' and formula = 'y ~ x'

geom_smooth() also allows for a linear regression line.

ggplot(mpg, mapping = aes(x = displ, y = hwy)) + 
  geom_point(color = 'steelblue') +
  geom_smooth(method = 'lm')

`geom_smooth()` using formula = 'y ~ x'

Mapping aesthetics is powerful for visualizing the relationship between multiple variables.

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

Mapping aesthetics allows us to represent categories with different colors and shapes.

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

Adding a theme can make the plot look better.

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

Adding labels makes the plot more descriptive.

ggplot(mpg, mapping = aes(x = displ, y = hwy, color = drv, shape = drv)) + 
  geom_point() + 
  theme_bw() +
  labs(x = 'Engine Displacement', 
       y = 'Highway Miles per Gallon',
       color = 'Drive Train', 
       shape = 'Drive Train')

Faceting divides the data into subsets according to a categorical variable and visualize each subset.

ggplot(mpg, mapping = aes(x = displ, y = hwy, color = drv, shape = drv)) + 
  geom_point() + 
  facet_wrap(~ drv) +
  theme_bw() +
  labs(x = 'Engine Displacement', 
       y = 'Highway Miles per Gallon',
       color = 'Drive Train',
       shape = 'Drive Train')

💻 Hands-On

Reproduce the following scatter plot:

Answer

ggplot(mpg, mapping = aes(x = displ, y = hwy, color = class)) + 
  geom_point() + 
  facet_wrap(~ class) +
  theme_bw() +
  labs(x = 'Engine Displacement', 
       y = 'Highway Miles per Gallon',
       color = 'Type of Car',
       shape = 'Type of Car')

Bar plot `geom_bar()`

Vertical bar plot can be made with geom_bar()

ggplot(mpg, mapping = aes(x = drv)) +
  geom_bar()

Changing the mapping can create horizontal bar plot.

ggplot(mpg, mapping = aes(y = drv)) +
  geom_bar()

Unlike geom_point(), color is adjusted with fill rather than color.

ggplot(mpg, mapping = aes(x = drv)) +
  geom_bar(fill = 'tomato')

Stacked bar plot can be obtained as follows:

ggplot(mpg, mapping = aes(x = drv, fill = class)) +
  geom_bar()

Side-by-side bar plot can be obtained as follows:

ggplot(mpg, mapping = aes(x = drv, fill = class)) +
  geom_bar(position = 'dodge')

Sequential, diverging and qualitative color scales from ColorBrewer can improve the bar plots.

ggplot(mpg, mapping = aes(x = drv, fill = class)) +
  geom_bar(position = 'dodge') +
  scale_fill_brewer(palette = 'Set1')

ggplot(mpg, mapping = aes(x = drv, fill = class)) +
  geom_bar(position = 'dodge') +
  scale_fill_brewer(palette = 'Spectral')

ggplot(mpg, mapping = aes(x = drv, fill = class)) +
  geom_bar(position = 'dodge') +
  scale_fill_brewer(palette = 'Pastel1')

Gray color scale is also available.

ggplot(mpg, mapping = aes(x = drv, fill = class)) +
  geom_bar(position = 'dodge') +
  scale_fill_grey()

One can also specify their own color scale. Refer to (https://colorbrewer2.org/#type=sequential&scheme=BuGn&n=3) for color scales that are colorblind safe, print friendly, and/or photocopy safe.

ggplot(mpg, mapping = aes(x = drv, fill = drv)) +
  geom_bar() +
  scale_fill_manual(values = c('#e5f5f9', '#99d8c9', '#2ca25f'))

💻 Hands-On

Reproduce the following bar plot. Note that the color is burlywood2.

Answer

ggplot(mpg, mapping = aes(x = class)) +
  geom_bar(fill = 'burlywood2') +
  theme_bw() +
  labs(x = 'Drive Train', 
       y = 'Frequency')

Histogram `geom_histogram()` and density curve `geom_density()`

Histogram can be made with geom_histogram().

ggplot(mpg, mapping = aes(x = hwy)) +
  geom_histogram(fill = 'chartreuse3')

Overlaying histograms can be used to compare different groups.

ggplot(mpg, mapping = aes(x = hwy, fill = drv)) +
  geom_histogram()

However, it can be more informative to use faceting in the case of histogram.

ggplot(mpg, mapping = aes(x = hwy, fill = drv)) +
  geom_histogram() +
  facet_wrap(~ drv)

Density is the “smoothed” version of histogram.

ggplot(mpg, mapping = aes(x = hwy)) +
  geom_density(fill = 'chartreuse3')

Overlaying density curves are also possible to make.

ggplot(mpg, mapping = aes(x = hwy, fill = drv)) +
  geom_density()

If we prefer non-overlapping density curves, we can consider faceting.

ggplot(mpg, mapping = aes(x = hwy, fill = drv)) +
  geom_density() +
  facet_wrap(~ drv)

💻 Hands-On

Reproduce the following histogram:

Answer

ggplot(mpg, mapping = aes(x = hwy, fill = class)) +
  geom_histogram() +
  facet_wrap(~ class) +
  theme_bw() +
  labs(x = 'Highway Miles per Gallon',
       y = 'Frequency',
       fill = 'Type of Car')

`stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

Box plot `geom_boxplot()` and violin plot `geom_violin()`

Box plot can be made with geom_boxplot().

ggplot(mpg, mapping = aes(y = hwy)) +
  geom_boxplot(fill = 'gold')

Side-by-side box plots can be made using mapping.

ggplot(mpg, mapping = aes(y = hwy, fill = drv)) +
  geom_boxplot()

A violin plot is a good combination of histogram and box plot.

ggplot(mpg, mapping = aes(y = hwy, x = drv, fill = drv)) +
  geom_violin()

💻 Hands-On

Reproduce the following box plots:

Answer

ggplot(mpg, mapping = aes(y = hwy, x = class, fill = class)) +
  geom_boxplot() +
  theme_bw() +
  labs(y = 'Highway Miles per Gallon',
       x = 'Type of Car',
       fill = 'Type of Car')

Readings

Topics

The tidyverse

Tidy data

Tibble

Fuel economy data mpg

ggplot2 and the grammar of graphic

Scatter plot geom_point()

Bar plot geom_bar()

Histogram geom_histogram() and density curve geom_density()

Box plot geom_boxplot() and violin plot geom_violin()

The `tidyverse`

Fuel economy data `mpg`

`ggplot2` and the grammar of graphic

Scatter plot `geom_point()`

Bar plot `geom_bar()`

Histogram `geom_histogram()` and density curve `geom_density()`

Box plot `geom_boxplot()` and violin plot `geom_violin()`