purrr also includes a pipe operator: %>%. The pipe operator is another shortcut that saves typing, but also increases readability. The explanation of the pipe operator is quite simple: x %>% f(y) is another way of writing f(x, y). That is, the left hand side of the pipe, x, becomes the first argument to the function, f(), on the right hand side of the pipe.

Take a look at our code to get our list of models:

cyl <- split(mtcars, mtcars$cyl) 
map(cyl, ~ lm(mpg ~ wt, data = .))

We split the data frame mtcars and save it as the variable cyl. We then pass cyl as the first argument to map to fit the models. We could rewrite this using the pipe operator as:

split(mtcars, mtcars$cyl) %>% 
  map(~ lm(mpg ~ wt, data = .))

We read this as "split the data frame mtcars on cyl, then use map() on the result."

One of the powerful things about the pipe is we can chain together many operations. Here is our complete code, written with pipes, instead assigning each step to a variable and using it in the next step:

mtcars %>% 
  split(mtcars$cyl) %>%
  map(~ lm(mpg ~ wt, data = .)) %>%
  map(coef) %>% 
  map_dbl("wt")

We've written some code in the editor to pull out the \(R^2\) from each model. Rewrite the last two lines to use a pipe instead.

Introducing purrr