37 Counting
Written by Annie Collins and last updated on 7 October 2021.
37.1 Introduction
In this lesson, you will learn how to:
- Use
count()to summarize a data set. - Use
uncount()to produce a full data set from a table containing values and counts.
This lesson is a yellow level skill and is part of “Tidyverse Essentials.” Prerequisite skills include:
- Installing packages
- Calling libraries
- Importing data
37.3 count()
count() is a function that can be used to summarize a data set in terms of the number of unique values of a certain variable or combination of variables.
Using count() on a data set produces a new table including each unique value or combination of values along with an integer value representing the number of times it appears in the data.
37.4 Using count()
The data set below (starwars) contains information on characters in the the Star Wars film franchise. There are 30 unique characters in the data with five characteristics/measurements recorded for each.
#> # A tibble: 30 × 6
#> name hair_color skin_color eye_color homeworld species
#> <chr> <chr> <chr> <chr> <chr> <chr>
#> 1 Luke S… blond fair blue Tatooine Human
#> 2 C-3PO <NA> gold yellow Tatooine Droid
#> 3 R2-D2 <NA> white, bl… red Naboo Droid
#> 4 Darth … none white yellow Tatooine Human
#> 5 Leia O… brown light brown Alderaan Human
#> 6 Owen L… brown, gr… light blue Tatooine Human
#> 7 Beru W… brown light blue Tatooine Human
#> 8 R5-D4 <NA> white, red red Tatooine Droid
#> 9 Biggs … black light brown Tatooine Human
#> 10 Anakin… blond fair blue Tatooine Human
#> # … with 20 more rowsAlthough each character is unique and only appears once in the data, there are several variables where multiple characters have the same characteristic, and we may want to know how many characters share each characteristic.
For example, we may want to examine how many characters came from each home world.
count(starwars, homeworld)
#> # A tibble: 5 × 2
#> homeworld n
#> <chr> <int>
#> 1 Alderaan 3
#> 2 Coruscant 3
#> 3 Kamino 3
#> 4 Naboo 11
#> 5 Tatooine 10We can also sort this table to see which home worlds appear in the data most frequently.
count(starwars, homeworld, sort=TRUE)
#> # A tibble: 5 × 2
#> homeworld n
#> <chr> <int>
#> 1 Naboo 11
#> 2 Tatooine 10
#> 3 Alderaan 3
#> 4 Coruscant 3
#> 5 Kamino 3Here, we see that Naboo is the most common home world, with 11 characters in the data set from this world.
If we wish to get even more specific, we could look at unique species-homeworld combinations amongst characters in the data.
count(starwars, species, homeworld, sort=TRUE)
#> # A tibble: 11 × 3
#> species homeworld n
#> <chr> <chr> <int>
#> 1 Human Tatooine 8
#> 2 Human Naboo 5
#> 3 Gungan Naboo 3
#> 4 Human Alderaan 3
#> 5 Droid Tatooine 2
#> 6 Human Coruscant 2
#> 7 Kaminoan Kamino 2
#> 8 <NA> Naboo 2
#> 9 Droid Naboo 1
#> 10 Human Kamino 1
#> 11 Tholothian Coruscant 1The outputted table shows that there are eight characters in the starwars data set that are humans from Tatooine, five that are humans from Naboo, and so on.
37.5 count() Arguments
count(data, variables, sort = FALSE, name = NULL)
- data: the data frame you are counting from.
-
variables: the variables in data you are counting, inputted by individual column name. This could be one or multiple variables to be counted in combination. If none are entered,
count()will return a count of the total number of rows in data. -
sort = …: FALSE by default (if omitted). If sort = TRUE,
count()will sort the outputted table in descending order of the count column, otherwise it will be sorted in alphabetical order by the first column (with NA values last). - name = “…”: NULL (empty) by default. This argument indicates the name you wish to give the column containing the count values, inputted in the form of a string. If omitted, the count column will be titled n.
37.6 uncount()
uncount() is a function that can be used to produce a full length table from a table containing variables and counts (for example, one that was produced using the count() function). The function duplicates each unique variable according to the value in the count column indicated. You may think of it as the inverse of count().
37.6.1 Arguments
uncount() takes two required arguments and two optional arguments:
uncount(data, weights, .remove = TRUE, .id = NULL)
- data: the data frame containing variables and counts.
- weights: the name of the column in data containing the counts for each variable(s). You can also use a constant or expression if you wish all variables to be duplicated the same number of times, or if an appropriate column does not exist in data.
Optional:
- .remove = …: TRUE by default. If weights is a column in data, this argument indicates whether or not you want the weights values to remain in the outputted table. If .remove = TRUE (default), the outputted table will only contain the duplicated variables. If .return = FALSE, the outputted table will contain the duplicated variables along with their total count present in an additional column. If weights is a value or expression, then .remove does not apply and the column containing counts (if it exists) will be included in the outputted table.
- .id = “…”: NULL (empty) by default. This argument allows you to input a string as the title for a new column that will automatically generate a unique ID for each duplicated variable. This may be useful if you still want to distinguish between variables once they have been repeated.
37.7 Using uncount()
To help visualize the use of uncount(), we will use this small data set called colours.
#> colour n
#> 1 red 1
#> 2 yellow 2
#> 3 blue 3colours indicates that red occurs once, yellow occurs twice, and blue occurs three times in some data, for a total of six observations.
If we want to observe all the data in its uncounted form, we can use uncount().
uncount(colours, n)
#> colour
#> 1 red
#> 2 yellow
#> 3 yellow
#> 4 blue
#> 5 blue
#> 6 blueOur output includes only one column (no count values) with each colour repeated the number of times indicated in the n column in colours. uncount() also automatically generates a new version of the row number according to the number of times each row was duplicated.
If we wish to duplicate each colour (regardless of the value in column n), we can substitute 2 in place of n in the function. Note that in the outputted table, the values from n are still present since we did not input a column for the weights argument.
uncount(colours, 2)
#> colour n
#> 1 red 1
#> 2 red 1
#> 3 yellow 2
#> 4 yellow 2
#> 5 blue 3
#> 6 blue 3Using the optional arguments .remove and .id, we can uncount colours according to column n while still keeping the values from n in the output and assign a unique ID number to each repeated colour.
uncount(colours, n, .remove = FALSE, .id = "id")
#> colour n id
#> 1 red 1 1
#> 2 yellow 2 1
#> 3 yellow 2 2
#> 4 blue 3 1
#> 5 blue 3 2
#> 6 blue 3 3Now we can distinguish between different yellow and blue rows in the column (for instance, yellow 1 and yellow 2 according to the id column).
Note that this does not change if there are multiple combinations of variables in colours we wish to uncount.
Updated colours:
#> colour temp n
#> 1 red warm 1
#> 2 yellow warm 2
#> 3 blue cool 3
uncount(colours, n, .remove = FALSE, .id = "id")
#> colour temp n id
#> 1 red warm 1 1
#> 2 yellow warm 2 1
#> 3 yellow warm 2 2
#> 4 blue cool 3 1
#> 5 blue cool 3 2
#> 6 blue cool 3 337.8 Exercises
Below is the first five rows of output from Exercise 1, assigned the name features. Please use this data set as reference for the following questions.
#> # A tibble: 5 × 3
#> skin_color eye_color n
#> <chr> <chr> <int>
#> 1 fair blue 6
#> 2 light brown 6
#> 3 dark brown 2
#> 4 fair brown 2
#> 5 grey black 237.8.1 Question 1
In the output of uncount(features, n), how many times will the word “brown” appear?
- 2
- 3
- 7
- 12
37.8.2 Question 2
How many rows does the output of uncount(features, 2) have (not including the header)?
- 5
- 10
- 21
- 42
37.8.3 Question 3
How many columns does the output of uncount(features, 2) have?
- 1
- 2
- 3
- 4
For the Question 4 and 5, identify the code used to manipulate the starwars data set to match the output displayed. The original is included below for reference.
#> # A tibble: 30 × 6
#> name hair_color skin_color eye_color homeworld species
#> <chr> <chr> <chr> <chr> <chr> <chr>
#> 1 Luke S… blond fair blue Tatooine Human
#> 2 C-3PO <NA> gold yellow Tatooine Droid
#> 3 R2-D2 <NA> white, bl… red Naboo Droid
#> 4 Darth … none white yellow Tatooine Human
#> 5 Leia O… brown light brown Alderaan Human
#> 6 Owen L… brown, gr… light blue Tatooine Human
#> 7 Beru W… brown light blue Tatooine Human
#> 8 R5-D4 <NA> white, red red Tatooine Droid
#> 9 Biggs … black light brown Tatooine Human
#> 10 Anakin… blond fair blue Tatooine Human
#> # … with 20 more rows37.8.4 Question 4
#> # A tibble: 16 × 3
#> skin_color eye_color n
#> <chr> <chr> <int>
#> 1 fair blue 6
#> 2 light brown 6
#> 3 dark brown 2
#> 4 fair brown 2
#> 5 grey black 2
#> 6 light blue 2
#> 7 dark blue 1
#> 8 gold yellow 1
#> 9 green orange 1
#> 10 grey orange 1
#> 11 orange orange 1
#> 12 pale yellow 1
#> 13 tan brown 1
#> 14 white yellow 1
#> 15 white, blue red 1
#> 16 white, red red 1count(starwars, skin_color, eye_color, sort=TRUE)count(starwars, skin_color, eye_color, sort=FALSE)count(starwars, c(skin_color, eye_color), sort=TRUE)count(starwars, c(skin_color, eye_color), sort=FALSE)
37.8.5 Question 5
#> # A tibble: 30 × 2
#> name unique
#> <chr> <int>
#> 1 Adi Gallia 1
#> 2 Anakin Skywalker 1
#> 3 Bail Prestor Organa 1
#> 4 Beru Whitesun lars 1
#> 5 Biggs Darklighter 1
#> 6 Boba Fett 1
#> 7 C-3PO 1
#> 8 Cliegg Lars 1
#> 9 Cordé 1
#> 10 Darth Vader 1
#> # … with 20 more rowscount(starwars)count(starwars, name, species, .remove = TRUE)count(starwars, name, name="name")count(starwars, name, name="unique")
37.8.6 Question 6
What will be the output of the following code: count(starwars, skin_color) %>% uncount(weights = n)?
- The original
starwarsdata frame - A tibble with two columns: one called “skin_color” with each unique skin colour from
starwarsincluded once, and one called “n” with a count of the number of times that skin colour appears in the data set - A tibble with two columns: one called “skin_color” with each skin colour from
starwarsincluded the same number of times it appears in the data set, and one called “n” with a count of the number of times that skin colour appears in the data set (repeated for each repeated entry in column “skin_color”) - A tibble with one column called “skin_color” with each skin colour from
starwarsincluded the same number of times it appears in the data set.
37.8.7 Question 7
Use the following tibble for Questions 7 through 9:
#> # A tibble: 5 × 3
#> skin_color species id
#> <chr> <chr> <int>
#> 1 dark Human 1
#> 2 dark Tholothian 1
#> 3 dark <NA> 1
#> 4 fair Human 1
#> 5 fair Human 2The tibble above represents the product of manipulating the starwars data set using one or more functions. Which of the options below best describes the last function applied in order to produce the given output?
count()uncount()summarise()- Something else