How to a run function per group that return vector, not single value?

Question

I have a dataset data.frame(x=rnorm(100),group=c(rep(\'a\',40),rep(\'b\',60))) that I want to analyse per group with dplyr. For example I want to use a fft

Answer 1

We can do with unnest after creating a list output with summarise. It would be more easier to work with

library(tidyverse)
df1 %>%
    group_by(group) %>%
    summarise(value = list(fft(x))) %>%
    unnest()

Answer 2

There are essentially three options:

1. If you want to create a single summary value per group, use summarize.
2. If you want to transform each value per group, use mutate.
3. If you want to create a new table per group, use do.

The last option seems to fit your purpose best, if I understood you correctly. do is generally the most powerful of these options, but also the hardest to use. The general syntax is:

data %>%
    group_by(grouping_cols) %>%
    do(data_frame(col1 = some_transformation(.$x)))

For example:

iris %>%
    group_by(Species) %>%
    do(broom::tidy(lm(Sepal.Length ~ Sepal.Width, data = .)))