dplyr error with summarise_ and n()

Question

I plan to use a loop to go through a few 100 columns and calculate a median and count against a set of variables.

Code :

grp_var <- "Species" 
voi <- "Sepal.Length"
dmp <- iris%>%
      select_(grp_var,voi)%>% 
      group_by_(grp_var)%>%
      summarise_(Median_Value = median(voi),Count = n())

Error :

Error in n() : This function should not be called directly

I face this error when i use summarise_ and don't face any issues at all when i use summarise.

I know that the function call to n() is just the error message but when called within a dplyr function it should return the row count. Am i just being dumb about this or is this a bug ?

Edit : I don't have any conflicts for the summarise function , plyr isn't loaded.

https://stackoverflow.com/questions/22801153/dplyr-error-in-n-function-should-not-be-called-directly — M--, Aug 11 '17 at 21:07
looks like you have the plyr package loaded. try `detach("package:plyr")` first — Richard Telford, Aug 11 '17 at 21:08
I have checked for conflicts and there aren't any. I haven't loaded plyr package either. — Abhishek Vij, Aug 11 '17 at 21:24
The problem is mixing standard and non-standard evaluation inside `summarise_`. The solution is covered in keiku's answer at the [question that keeps getting linked](https://stackoverflow.com/a/42716540/903061). — Gregor Thomas, Aug 11 '17 at 21:46
Since `dplyr` is now in version >= 0.7, tidyeval is now the new concept. all `*_` function are now deprecated. linked answer is outdated for current `dplyr` — cderv, Aug 11 '17 at 22:02
Yes true. However I think it shouldn't be used anymore as it will disappear. — cderv, Aug 12 '17 at 06:10

cderv · Accepted Answer · 2017-08-12T06:22:39.503

First, as explained in the comments, you mixed standard evaluation and non-standard evaluation. n() is not found because you can't use it like that in *_ functions. In dplyr before 0.7.0, you would use ~n() in summarise_.

However things have changed in the tidyverse world.

Since version 0.7.0, dplyr uses now a new system for programming with dplyr, called tidy evaluation, or tidy eval for short. All function with *_ are now deprecated and should not be used in new code, unless you want to keep a dependency on an old dplyr version. I'll advice to use tidy eval now. I will not explained it here, you could see the Programming vignette

For example, now you would do something like this with dplyr (>= 0.7.0):

library(dplyr)
# quo is a tidy eval concept for quoting
grp_var <-quo(Species)
voi <- quo(Sepal.Length)
# use !! another tidy eval concept to unquote
dmp <- iris %>%
  select(!! grp_var, !! voi) %>% 
  group_by(!! grp_var) %>%
  summarise(Median_Value = median( !! voi ), Count = n())
dmp
#> # A tibble: 3 x 3
#>      Species Median_Value Count
#>       <fctr>        <dbl> <int>
#> 1     setosa          5.0    50
#> 2 versicolor          5.9    50
#> 3  virginica          6.5    50

dplyr error with summarise_ and n()

1 Answers1