Nesting in R

A list column of data frames, one per group — the tidy way to model per group.

library(tidyr)
library(dplyr)
library(purrr)

runs <- tibble(
  lang = c("r", "r", "r", "python", "python", "python"),
  day = c(1, 2, 3, 1, 2, 3),
  tpm = c(88, 92, 96, 99, 104, 101)
)

nested <- nest(runs, data = c(day, tpm))
print(nested)

fitted <- mutate(nested,
                 rows = map_int(data, nrow),
                 slope = map_dbl(data, ~ coef(lm(tpm ~ day, data = .x))[[2]]))
print(select(fitted, lang, rows, slope))
print(unnest(nested, data) |> nrow())

How it works

1. nest(.by = ) puts each group's rows in a data column.
2. map over that column fits or summarizes per group.
3. unnest flattens it back out.

Keywords and builtins used here

• c
• coef
• library
• lm
• map_dbl
• map_int
• mutate
• nest
• nrow
• print
• select
• tibble
• unnest

The run, in numbers

Lines

18

Characters to type

430

Tokens

158

Three-star pace

105 tpm

At the three-star pace of 105 tokens a minute, this run takes about 90 seconds.

Type this snippet

Step 1 of 3 in Nesting & gaps, step 6 of 12 in Reshaping & reading.

← Previous Next →

Nesting in other languages

• CSS