dplyr::recode Why does pipe generate error?

Question

If I use recode in a pipe, I get an error:

df <-  df %>%
  recode(unit, .missing="g")

Error in UseMethod("recode") : no applicable method for 'recode' applied to an object of class "c('tbl_df', 'tbl', 'data.frame')"

If I pull it out of the pipe, it works fine:

df$unit <- recode(df$unit, .missing="g")

Any ideas why? I'd like to stay in the pipe if possible.

Answer 1

An equivalent of the baseR solution in dplyr is to use it inside mutate:

df %>%
    mutate(unit = recode(unit, .missing="g"))

Directly chaining recode after %>% will pass the data frame to recode as the first argument, which doesn't agree with recode's parameters. The first argument .x needs to be a vector; unlike some other dplyr functions recode doesn't use some non-standard evaluation magic to interpret unit as the column with that name in df. Most functions designed for direct use with the pipe have a data frame as their first argument and their output. You can read more about magrittr and how the pipe works here.