`slice()` behaves differently to `head()` on nested tibbles vs. unnested tibbles.

[cynthiahqy]

slice() should return the same as head() when extracting the first row of a table. For normal tibbles it does so, but when working with nested tibbles, only head(1) works as expected (returning the first row), while slice(1) returns all rows, and slice(2) returns no rows.

I've included a reprex of the unexpected behaviour with dplyr functions, but I suspect the issue is to do with the behaviour of nested tibbles so I've raised the issue here.

library(dplyr, quietly = TRUE, warn.conflicts = FALSE)
# Normal tibble
## these two are equivalent
identical(head(mtcars, 1), slice(mtcars, 1))
#> [1] TRUE

# Nested Tibble
## gets the first row as expected
mtcars |> nest_by(cyl) |> head(1)
#> # A tibble: 1 × 2
#> # Rowwise:  cyl
#>     cyl                data
#>   <dbl> <list<tibble[,10]>>
#> 1     4           [11 × 10]
## returns all rows unexpectedly
mtcars |> nest_by(cyl) |> slice(1)
#> # A tibble: 3 × 2
#> # Rowwise:  cyl
#>     cyl                data
#>   <dbl> <list<tibble[,10]>>
#> 1     4           [11 × 10]
#> 2     6            [7 × 10]
#> 3     8           [14 × 10]
## returns no rows unexpectedly
mtcars |> nest_by(cyl) |> slice(2)
#> # A tibble: 0 × 2
#> # Rowwise:  cyl
#> # ℹ 2 variables: cyl <dbl>, data <list<tibble[,10]>>

^{Created on 2024-06-12 with reprex v2.0.2}

[DavisVaughan]

slice() is applied per group, and nest_by() causes your data frame to become a rowwise data frame, i.e. 1 group per row.

I tend to prefer tidyr::nest(.by =) instead of dplyr::nest_by() as it won't turn the result into a rowwise data frame. We now mostly think that returning a rowwise data frame from nest_by() was a mistake, as it can result in confusing situations like this.

library(dplyr, quietly = TRUE, warn.conflicts = FALSE)
library(tidyr)

mtcars |> nest(.by = cyl) |> slice(1)
#> # A tibble: 1 × 2
#>     cyl data             
#>   <dbl> <list>           
#> 1     6 <tibble [7 × 10]>

^{Created on 2024-06-12 with reprex v2.0.2}

[cynthiahqy]

Thanks for the explanation! I'll use the tidyr function as recommended.

…

On Wed, 12 Jun 2024 at 14:52, Davis Vaughan ***@***.***> wrote: slice() is applied *per group*, and nest_by() causes your data frame to become a rowwise data frame, i.e. 1 group per row. I tend to prefer tidyr::nest(.by =) instead of dplyr::nest_by() as it won't turn the result into a rowwise data frame. We now mostly think that returning a rowwise data frame from nest_by() was a mistake, as it can result in confusing situations like this. library(dplyr, quietly = TRUE, warn.conflicts = FALSE) library(tidyr) mtcars |> nest(.by = cyl) |> slice(1)#> # A tibble: 1 × 2#> cyl data #> <dbl> <list> #> 1 6 <tibble [7 × 10]> Created on 2024-06-12 with reprex v2.0.2 <https://reprex.tidyverse.org> — Reply to this email directly, view it on GitHub <#1553 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AHCXTQ2SPR6FAQRJFBY2DLLZHBAAXAVCNFSM6AAAAABJGGQGBKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNRSHEZTIOBQGQ> . You are receiving this because you authored the thread.Message ID: ***@***.***>