A Basic Scatter Plot with geom_point

The first example will use the builtin mtcars dataset to show off a simple scatterplot.

library("tidyverse")

## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr     1.1.3     ✔ readr     2.1.5
## ✔ forcats   1.0.0     ✔ stringr   1.5.0
## ✔ ggplot2   3.5.0     ✔ tibble    3.2.1
## ✔ lubridate 1.9.3     ✔ tidyr     1.3.0
## ✔ purrr     1.0.2     
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag()    masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors

#this library makes the output look a little better
#if it's not installed yet call
#install.packages("pander")
library("pander")

#load the data
data(mtcars)

#look at the first few rows
#mtcars %>% head() %>% pander()

#here's another way to do it
pander(head(mtcars))

Let’s look at the relationship between mpg and hp with a simple scatterplot using geom_point().

Remember these three key points to using ggplot2.

1. the data
2. the geom
3. the mapping from data to geom.

ggplot(mtcars, aes(x=mpg, y=hp)) +
    geom_point()

aesthetics

We can assign variables to be visualized through the geom’s features. In the code below, the gear of the car is shown off as color. gear is being treated as a numeric and ranges over 3 distinct values.

ggplot(mtcars, aes(x=mpg, y=hp, color=gear)) +
    geom_point()

By default, gear is stored as a floating point number. If you change its class to a factor, ggplot will treat it as a categorical variable, and assign attributes based on a discrete scale rather than continuous.

#turn gear into a factor
mtcars$gear2 = as.factor(mtcars$gear)
#Here's another way to do it
mtcars <- mtcars %>% mutate(gear2 = as.factor(gear))

pander(head(mtcars))

ggplot(mtcars, aes(x=mpg, y=hp, color=gear2)) +
    geom_point()

Displaying Heatmaps with geom_tile

In these examples we’ll be using the builtin starwars dataset to show off how to make a heatmap in ggplot2. If you’ve made heatmaps in base R with the image() function, you’re already familiar with how to display data in matrix form. This is a very intuitive way to handle 2D data.

ggplot2 uses a different way to handle 2D data for heatmaps. Here, we treat each individual heatmap tile as a separate observation. Instead of a rectangular matrix with nSubjects rows and nVariables columns, we need to make a new data structure that has nSubjects * nVariables rows, with one or more columns to distinguish the variables and their values.

Luckily for us, there is a tidyverse function called unnest() that can rearrange the data for us automatically.

For the first example, let’s look at the films that each character had an appearance in using geom_tile()

data("starwars")
head(starwars$name)

## [1] "Luke Skywalker" "C-3PO"          "R2-D2"          "Darth Vader"   
## [5] "Leia Organa"    "Owen Lars"

head(starwars$films)

## [[1]]
## [1] "The Empire Strikes Back" "Revenge of the Sith"    
## [3] "Return of the Jedi"      "A New Hope"             
## [5] "The Force Awakens"      
## 
## [[2]]
## [1] "The Empire Strikes Back" "Attack of the Clones"   
## [3] "The Phantom Menace"      "Revenge of the Sith"    
## [5] "Return of the Jedi"      "A New Hope"             
## 
## [[3]]
## [1] "The Empire Strikes Back" "Attack of the Clones"   
## [3] "The Phantom Menace"      "Revenge of the Sith"    
## [5] "Return of the Jedi"      "A New Hope"             
## [7] "The Force Awakens"      
## 
## [[4]]
## [1] "The Empire Strikes Back" "Revenge of the Sith"    
## [3] "Return of the Jedi"      "A New Hope"             
## 
## [[5]]
## [1] "The Empire Strikes Back" "Revenge of the Sith"    
## [3] "Return of the Jedi"      "A New Hope"             
## [5] "The Force Awakens"      
## 
## [[6]]
## [1] "Attack of the Clones" "Revenge of the Sith"  "A New Hope"

#This is the original form of the data
starwars %>% select(name, films) %>% head() %>% pander()

#this is how it needs to be to use ggplot's heatmap
starwars_films <- starwars %>% select(name, films) %>% unnest(films)
starwars_films %>% head(30) %>% pander()

#now show the heatmap
ggplot(starwars_films, aes(x=films, y=name)) +
    geom_tile()

Let’s look at another heatmap. The types of starships piloted by each character, broken down by the pilot’s species. Here, we’re rearranging the necessary info into a tiled data structure, then we’re adding extra info from the original data with left_join()

#make a new data.frame that has the info we want to visualize
starwars_ships <- starwars %>% select(name, starships) %>% unnest(starships)
pander(starwars_ships) 

#add info from the original data.frame to the new data.frame
starwars_shipsSpecies <- left_join(starwars_ships, select(starwars, name, species), by="name")
pander(starwars_shipsSpecies)

#color the heatmap elements by species
ggplot(starwars_shipsSpecies, aes(x=starships, y=name, fill=species)) +
    geom_tile() + 
    theme(axis.text.x = element_text(angle = 90))

Exploring the Olympics Dataset

Here are a few more ggplot2 examples using the olympics dataset.
We’ll be looking for associations between the number/type of medals won and other interesting variables.

events <- read_csv("athlete_events.csv")

## Rows: 271116 Columns: 15
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (10): Name, Sex, Team, NOC, Games, Season, City, Sport, Event, Medal
## dbl  (5): ID, Age, Height, Weight, Year
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

This dataset contains 271116 rows and 15 columns. To make things more manageable, we’ll be restricting our exploration to only the most popular games and teams.

Question 1 (review)

How do you find the sports & teams with the largest number of players?

#sports with the most number of players
events  %>% count(Sport) %>% arrange(desc(n)) %>% top_n(n=10, wt=n) 

## # A tibble: 10 × 2
##    Sport                    n
##    <chr>                <int>
##  1 Athletics            38624
##  2 Gymnastics           26707
##  3 Swimming             23195
##  4 Shooting             11448
##  5 Cycling              10859
##  6 Fencing              10735
##  7 Rowing               10595
##  8 Cross Country Skiing  9133
##  9 Alpine Skiing         8829
## 10 Wrestling             7154

#from here you can extract the names from the Sport column
top_sports <- events  %>% 
  count(Sport) %>% 
  arrange(desc(n)) %>% 
  top_n(n=10, wt=n) %>%
  pull(Sport)

top_sports

##  [1] "Athletics"            "Gymnastics"           "Swimming"            
##  [4] "Shooting"             "Cycling"              "Fencing"             
##  [7] "Rowing"               "Cross Country Skiing" "Alpine Skiing"       
## [10] "Wrestling"

#another way to do it
top_sports <- names(tail(sort(table(events$Sport)), 10))
top_teams <- names(tail(sort(table(events$Team)), 10))

top_sports

##  [1] "Wrestling"            "Alpine Skiing"        "Cross Country Skiing"
##  [4] "Rowing"               "Fencing"              "Cycling"             
##  [7] "Shooting"             "Swimming"             "Gymnastics"          
## [10] "Athletics"

top_teams

##  [1] "Hungary"       "Australia"     "Sweden"        "Japan"        
##  [5] "Canada"        "Germany"       "Italy"         "Great Britain"
##  [9] "France"        "United States"

Question 2 (review)

Let’s also restrict ourselves to only using the players that won a medal. We want the players that won a medal in any of the top sports from any of the top teams. How do you set up this smaller dataset?

events_top <- events %>% 
    filter(!is.na(Medal )) %>%
    filter(Sport %in% top_sports, Team %in% top_teams)

events_top

## # A tibble: 9,881 × 15
##       ID Name     Sex     Age Height Weight Team  NOC   Games  Year Season City 
##    <dbl> <chr>    <chr> <dbl>  <dbl>  <dbl> <chr> <chr> <chr> <dbl> <chr>  <chr>
##  1    56 Ren Aba… M        21     NA     NA Fran… FRA   1956…  1956 Summer Melb…
##  2    62 Giovann… M        21    198     90 Italy ITA   2016…  2016 Summer Rio …
##  3    84 Stephen… M        26    165     55 Unit… USA   2004…  2004 Summer Athi…
##  4    99 Pter Ab… M        30    181     79 Hung… HUN   1992…  1992 Summer Barc…
##  5   100 Oszkr A… M        22     NA     NA Hung… HUN   1936…  1936 Summer Berl…
##  6   106 Agostin… M        22    188     96 Italy ITA   1988…  1988 Summer Seoul
##  7   106 Agostin… M        29    188     96 Italy ITA   1996…  1996 Summer Atla…
##  8   106 Agostin… M        34    188     96 Italy ITA   2000…  2000 Summer Sydn…
##  9   107 Carmine… M        22    182     90 Italy ITA   1984…  1984 Summer Los …
## 10   107 Carmine… M        26    182     90 Italy ITA   1988…  1988 Summer Seoul
## # ℹ 9,871 more rows
## # ℹ 3 more variables: Sport <chr>, Event <chr>, Medal <chr>

Question 3

Which team had the most number of medals? Show the results in a bar graph using geom_col.

pander(count(events_top, Team, Medal))

#bar plot of number of medals by team
count(events_top, Team, Medal) %>%
    ggplot(aes(x=Team, y=n, fill=Medal)) +
        geom_col()

#stack them side by side
count(events_top, Team, Medal) %>%
    ggplot(aes(x=Team, y=n, fill=Medal)) +
        geom_col(position="dodge")

#let's change the colors
count(events_top, Team, Medal) %>%
    ggplot(aes(x=Team, y=n, fill=Medal)) +
        geom_col(position="dodge") + 
        scale_fill_manual(values = c(
            "Gold" = "gold", 
            "Bronze" = "goldenrod", 
            "Silver" = "gray"))

#what if you want to order the teams by the total number of medals?
count(events_top, Team, Medal) %>%
    ggplot(aes(x=fct_rev(fct_reorder(Team, n)), y=n, fill=Medal)) +
        geom_col(position="dodge") + 
        scale_fill_manual(values = c(
            "Gold" = "gold", 
            "Bronze" = "goldenrod", 
            "Silver" = "gray"))

#Change the order, gold-silver-bronze
events_top$Medal <- factor(events_top$Medal, levels=c("Gold", "Silver", "Bronze"))
count(events_top, Team, Medal) %>%
    ggplot(aes(x=fct_rev(fct_reorder(Team, n)), y=n, fill=Medal)) +
        geom_col(position="dodge") + 
        scale_fill_manual(values = c(
            "Gold" = "gold", 
            "Bronze" = "goldenrod", 
            "Silver" = "gray")) +
        theme(axis.text.x = element_text(angle = 90)) +
        labs(x="Teams", y="Medal Count")

Question 4

Look at the total number of medals over the years for each team. Do you notice any weird patterns? Can you correct them?

#total number of medals over the years
events_top %>% count(Team, Year) 

## # A tibble: 296 × 3
##    Team       Year     n
##    <chr>     <dbl> <int>
##  1 Australia  1896     2
##  2 Australia  1900     5
##  3 Australia  1904     4
##  4 Australia  1906     3
##  5 Australia  1920     7
##  6 Australia  1924     9
##  7 Australia  1928     4
##  8 Australia  1932     5
##  9 Australia  1936     1
## 10 Australia  1948    16
## # ℹ 286 more rows

#look at the number of medals by team and year
events_top %>% 
    count(Team, Year) %>% 
    ggplot(aes(x=Year, y=n, color=Team)) +
        geom_point() +
        geom_line()

#another way to look at the number of medals by team and year
events_top %>% 
    count(Team, Year) %>% 
    ggplot(aes(x=Year, y=n, color=Team)) +
        geom_point() +
        geom_line() +
        facet_wrap(~Team)

The spiky pattern in the number of medals won through the years is because many of the top sports only exist in the summer olympics. Inconsistencies like are very hard to spot if you’re only reading the numbers instead of visualizing them.
Let’s see what it looks like if we restrict ourselves to the summer games.

#total number of medals over the years
events_top %>% 
    filter(Season == "Summer") %>%
    count(Team, Year) 

## # A tibble: 261 × 3
##    Team       Year     n
##    <chr>     <dbl> <int>
##  1 Australia  1896     2
##  2 Australia  1900     5
##  3 Australia  1904     4
##  4 Australia  1906     3
##  5 Australia  1920     7
##  6 Australia  1924     9
##  7 Australia  1928     4
##  8 Australia  1932     5
##  9 Australia  1936     1
## 10 Australia  1948    16
## # ℹ 251 more rows

#look at the number of medals by team and year
events_top %>% 
    filter(Season == "Summer") %>%
    count(Team, Year) %>% 
    ggplot(aes(x=Year, y=n, color=Team)) +
        geom_point() +
        geom_line() +
        facet_wrap(~Team)

Question 5

Do you see any patterns in the weights of medal winning participants over the years? Show the weight distributions with geom_histogram and geom_density_ridges

#distribution of weights by medal
ggplot(events_top, aes(x=Weight)) + geom_histogram()

## Warning: Removed 2902 rows containing non-finite outside the scale range
## (`stat_bin()`).

ggplot(events_top, aes(x=Weight, fill=Sex)) + geom_histogram()

## Warning: Removed 2902 rows containing non-finite outside the scale range
## (`stat_bin()`).

ggplot(events_top, aes(x=Weight, fill=Sex)) + geom_histogram() + facet_wrap(~Year)

## Warning: Removed 2902 rows containing non-finite outside the scale range
## (`stat_bin()`).

ggridges is another very useful package for displaying multiple histograms. It’s not part of tidyverse, but can be installed very easily and uses the standard ggplot2 calling form. Be sure to specify a categorical variable to the group aesthetic if you want to visualize distinct ridges.

#if it's not installed yet.
#install.packages("ggridges")
library("ggridges")

ggplot(events_top, aes(x=Weight, group=Year, y=Year)) + 
  geom_density_ridges() + 
  facet_wrap(~Sex, ncol=2)

## Warning: Removed 2902 rows containing non-finite outside the scale range
## (`stat_density_ridges()`).

#separate it by sport
ggplot(events_top, aes(x=Weight, group=Year, y=Year, fill=Sex)) + 
  geom_density_ridges() + 
  facet_wrap(Sport~Sex, ncol=4)

## Warning: Removed 2902 rows containing non-finite outside the scale range
## (`stat_density_ridges()`).