purrr and non-purrr code comparison

purrr.R

library(purrr)
library(dplyr)
library(broom)

mtcars %>%
  split(.$cyl) %>% # from base R
  map(~ lm(mpg ~ wt, data = .)) %>%
  map(summary) %>%
  map_dbl("r.squared")

unlist(lapply(unique(mtcars$cyl), function(x){
  dt  <- dplyr::filter(mtcars, cyl == x)
  fit <- lm(mpg ~ wt, data = dt)
  
  fit %>%
    broom::glance() %>%
    pull(r.squared) %>%
    setNames(x)
}))

In the one case map is obscuring a lot of operations. In the other, we have logically grouped commands and it is very clear what is being iterated on.