fastplyr (version 0.5.1)

f_nest_by: Create a subset of data for each group

Description

A faster nest_by().

Usage

f_nest_by(
  data,
  ...,
  .add = FALSE,
  .order = df_group_by_order_default(data),
  .by = NULL,
  .cols = NULL,
  .drop = df_group_by_drop_default(data)
)

Value

A row-wise grouped_df of the corresponding data of each group.

Arguments

data

data frame.

...

Variables to group by.

.add

Should groups be added to existing groups? Default is FALSE.

.order

Should groups be ordered? If FALSE groups will be ordered based on first-appearance.
Typically, setting order to FALSE is faster.

.by

(Optional). A selection of columns to group by for this operation. Columns are specified using tidyselect.

.cols

(Optional) alternative to ... that accepts a named character vector or numeric vector. If speed is an expensive resource, it is recommended to use this.

.drop

Should unused factor levels be dropped? Default is TRUE.

Examples

library(dplyr)
library(fastplyr)

# Stratified linear-model example

models <- iris %>%
  f_nest_by(Species) %>%
  mutate(model = list(lm(Sepal.Length ~ Petal.Width + Petal.Length, data = first(data))),
         summary = list(summary(first(model))),
         r_sq = first(summary)$r.squared)
models
models$summary

# dplyr's `nest_by()` is admittedly more convenient
# as it performs a double bracket subset `[[` on list elements for you
# which we have emulated by using `first()`

# `f_nest_by()` is faster when many groups are involved

models <- iris %>%
  nest_by(Species) %>%
  mutate(model = list(lm(Sepal.Length ~ Petal.Width + Petal.Length, data = data)),
         summary = list(summary(model)),
         r_sq = summary$r.squared)
models$summary

models$summary[[1]]