将列表列添加到 rowwise() dataframe

Question

I'm trying to understand the behaviour of dplyr::rowwise() 'd dataframes. Specifically whether I can apply dplyr::mutate() to them to create list columns that contain multiple objects.

In the example below I'm attempting to add a column to mtcars where each element/row is a list of length two, containing two lm models.

suppressPackageStartupMessages(library(dplyr))

## attempt to output two objects into a list
## -----> this does not work <-----
## why not?
mtcars %>%
  nest_by(cyl) %>%
  mutate(mods = list(
    mymod1 = list(lm(mpg ~ wt, data = data)),
    mymod2 = list(lm(disp ~ wt, data = data))
  ))
#> Error: Problem with `mutate()` input `mods`.
#> x Input `mods` can't be recycled to size 1.
#> ℹ Input `mods` is `list(...)`.
#> ℹ Input `mods` must be size 1, not 2.
#> ℹ Did you mean: `mods = list(list(...))` ?
#> ℹ The error occurred in row 1.

^{Created on 2021-06-06 by the reprex package (v1.0.0)}

I can't quite make sense of the error message. Could anyone please help to clarify what's going on here?

Here are some related calls that do work, that led me to expect that the call above would also work.

suppressPackageStartupMessages(library(dplyr))

## output one object
## this works
mtcars %>%
  nest_by(cyl) %>%
  mutate(mod = list(lm(mpg ~ wt, data = data)))
#> # A tibble: 3 x 3
#> # Rowwise:  cyl
#>     cyl                data mod   
#>   <dbl> <list<tbl_df[,10]>> <list>
#> 1     4           [11 × 10] <lm>  
#> 2     6            [7 × 10] <lm>  
#> 3     8           [14 × 10] <lm>

## output one object into a list
## this also works
mtcars %>%
  nest_by(cyl) %>%
  mutate(mod = list(
    mymod = list(lm(mpg ~ wt, data = data))
  ))
#> # A tibble: 3 x 3
#> # Rowwise:  cyl
#>     cyl                data mod         
#>   <dbl> <list<tbl_df[,10]>> <named list>
#> 1     4           [11 × 10] <list [1]>  
#> 2     6            [7 × 10] <list [1]>  
#> 3     8           [14 × 10] <list [1]>

^{Created on 2021-06-06 by the reprex package (v1.0.0)}

ps I understand that this can be achieved with other methods. This is mostly a learning opportunity:)

Answer 1

You can store a list of length 1 in a dataframe.

df <- data.frame(a = 1)
tmp <- list(mymod1 = lm(mpg ~ wt, data = mtcars))
length(tmp)
#[1] 1

df$b <- tmp

But you cannot store a list of length more than 1.

df <- data.frame(a = 1)
tmp <- list(mymod1 = lm(mpg ~ wt, data = mtcars), 
            mymod2 = lm(mpg ~ wt, data = mtcars))
length(tmp)
#[1] 2

df$b <- tmp

Error in $<-.data.frame ( *tmp* , b, value = list(mymod1 = list(coefficients = c( (Intercept) = 37.285126167342, : replacement has 2 rows, data has 1

For that again you need to make a list of length 1 which can be done as -

df <- data.frame(a = 1)
tmp <- list(list(mymod1 = lm(mpg ~ wt, data = mtcars), 
                 mymod2 = lm(mpg ~ wt, data = mtcars)))
length(tmp)
#[1] 1
df$b <- tmp

---

So for your example this will work -

library(dplyr)

mtcars %>%
  nest_by(cyl) %>%
  rowwise() %>%
  mutate(mods = list(list(mymod1 = list(lm(mpg ~ wt, data = data)),
                          mymod2 = list(lm(disp ~ wt, data = data))))) 

#    cyl                data mods            
#  <dbl> <list<tibble[,10]>> <list>          
#1     4           [11 × 10] <named list [2]>
#2     6            [7 × 10] <named list [2]>
#3     8           [14 × 10] <named list [2]>

which is also what the error message suggests to do.