7.1 Convert Between Names and Values
First, let’s create an artificial dataset which contains the weights of a sheep and a pig for years 2019, 2020, and 2021.
library(tibble)
animal <- rep(c("sheep","pig"), c(3,3))
year <- rep(2019:2021, 2)
weight <- c(110, 120, 140, NA, 300, 800)
animal_tidy <- tibble(animal, year, weight)
animal_tidy
#> # A tibble: 6 × 3
#> animal year weight
#> <chr> <int> <dbl>
#> 1 sheep 2019 110
#> 2 sheep 2020 120
#> 3 sheep 2021 140
#> 4 pig 2019 NA
#> 5 pig 2020 300
#> 6 pig 2021 800By checking the definition of tidy data, it is clear animal_tidy is indeed tidy. Let’s make it untidy.
7.1.1 Convert Values into Column Names
In animal_tidy, each row contains the year when the weight measurement was taken. Suppose we want to convert the year value into column names. You can use the pivot_wider() function in tidyr package, another member of the tidyverse package.
In pivot_wider(), you need to specify two arguments:
names_fromdenotes which column in the original tibble contains the values of the new column names, -values_fromdenotes which column in the original tibble contains the values for each cell in the new tibble.
The reason why the function is called pivot_wider() is due to the fact that it will create a wider dataset than the orginal one, containing more columns.
library(tidyr)
animal_wide <- animal_tidy %>% pivot_wider(names_from = year, values_from = weight)
animal_wide #untidy animal: wide
#> # A tibble: 2 × 4
#> animal `2019` `2020` `2021`
#> <chr> <dbl> <dbl> <dbl>
#> 1 sheep 110 120 140
#> 2 pig NA 300 800In animal_wide, we have the columns names 2019, 2020, and 2021 coming from the year variable, and the values 110, 120, 140, NA, 300, and 800 from the weight variable, both of which are in the original tibble animal_tidy. The animal_wide is clearly untidy, since neither the weight nor the year information is contained in a single column. Note that this data format is commonly encountered in practice.
As it is clear from the resulting tibble, the name weight is lost during the pivoting process, which is not desirable. Fortunately, you can add a prefix “weight” to the column names via an argument names_prefix in the pivot_wider() function.
animal_wide_weight <- animal_tidy %>% pivot_wider(names_from = year,
names_prefix = "weight",
values_from = weight)
animal_wide_weight
#> # A tibble: 2 × 4
#> animal weight2019 weight2020 weight2021
#> <chr> <dbl> <dbl> <dbl>
#> 1 sheep 110 120 140
#> 2 pig NA 300 8007.1.2 Convert Column Names into Values
Now, you will learn how to tidy animal_wide into a tidy data. To do this, you can use the pivot_longer() function to convert the columns names 2019, 2020, and 2021 into values of a variable, for example, year.
animal_wide %>%
pivot_longer(cols = -1,
names_to = "year",
values_to = "weight")
#> # A tibble: 6 × 3
#> animal year weight
#> <chr> <chr> <dbl>
#> 1 sheep 2019 110
#> 2 sheep 2020 120
#> 3 sheep 2021 140
#> 4 pig 2019 NA
#> 5 pig 2020 300
#> 6 pig 2021 800In pivot_longer(), cols specifies the column names that you want to convert from, which accept the same format as that in dplyr::select() introduced in Section 6.4. names_to specifies the variable name you want to use for the column names. Finally, values_to specifies the variable name for holding the values in the selected columns.
You can see that we have recovered the animal_tidy through the tidy process. To tidy animal_wide_weight, we can use the same function pivot_longer() along with the argument names_prefix as below.
animal_wide_weight %>%
pivot_longer(cols = -1,
names_to = "year",
names_prefix = "weight",
values_to = "weight")
#> # A tibble: 6 × 3
#> animal year weight
#> <chr> <chr> <dbl>
#> 1 sheep 2019 110
#> 2 sheep 2020 120
#> 3 sheep 2021 140
#> 4 pig 2019 NA
#> 5 pig 2020 300
#> 6 pig 2021 800In this regards, pivot_wider() and pivot_longer() can be viewed as opposite functions.