Error in subsetting with $ immediately after a function in dplyr pipe

Question

I could subset a single column with the following syntax for functions that return data.frame or list:

library(dplyr)
filter(mtcars, disp > 400)$mpg
# [1] 10.4 10.4 14.7

But this causes the following error when used in a pipe (%>%):

mtcars %>% filter(disp > 400)$mpg
# Error in .$filter(disp > 400) : 
#   3 arguments passed to '$' which requires 2

I would like to know why $ does not work when used in pipe like the above example.

There is also the `pull` function of dplyr which you can use to extract a column as a vector — Sotos, Aug 06 '17 at 08:08
I'm using `pull` as well recently, but before I used to do `(mtcars %>% filter(disp > 400))$mpg` — AntoniosK, Aug 06 '17 at 09:19
Since this is a pipe, why can't `mtcars %>% filter(disp > 400) %>% .$mpg` this be done — akrun, Aug 06 '17 at 09:50

mt1022 · Accepted Answer · 2017-08-06T12:26:20.403

I think I have figured out the reason.

When I call filter(mtcars, disp > 400)$mpg, what actually happen is:

`$`(filter(mtcars, disp > 400), mpg)
# [1] 10.4 10.4 14.7

Similarly, mtcars %>% filter(disp > 400)$mpg is interpreted as:

`$`(mtcars, filter(disp > 400), mpg)

, because lhs of %>% will be the first argument of the function at rhs. This reproduced the same error that $ requires 2 args while 3 were supplied.

# Error in mtcars$filter(disp > 400) : 
#   3 arguments passed to '$' which requires 2

The error message also verified the above speculation. mtcars is used as the data.frame name and filter(disp > 400)is treated as a column name: mtcars$filter(disp > 400).

Error in subsetting with $ immediately after a function in dplyr pipe

1 Answers1