使用dplyr时如何在涉及基R函数的赋值运算符的RHS中使用UQ？

Question

I would like to use variables with varying names (ie names defined by using arguments of the function) in the calculations on RHS of assignment operators. Base R functions ( min , max , etc.) interpret the result of the unquoting operand !! as string, which is not what I want. How should I specify that R should work with variables names, not strings. No problem when using !! with dplyr verbs, such as select , mutate , etc.

df <- tibble(
  g1 = c(1, 1, 2, 2, 2),
  g2 = c(1, 2, 1, 2, 1),
  a = sample(5),
  b = sample(5)
)

my_s_mutate <- function(df, group_var, expr) {

  group_var <- enquo(group_var)
  expr <- enquo(expr)
  mean_name <- paste0("mean_", quo_name(expr))
  sum_name <- paste0("sum_", quo_name(expr))

  df %>%
    group_by(!! group_var) %>%
    mutate(
         !! mean_name := mean(!! expr),
         !! sum_name := sum(!! expr),
         diff = !! sum_name - !! mean_name
  )
}

my_s_mutate(df, g1, a)

This code give error because is evaluating the difference between two strings. Similar problem occurs when using min or max

 Error in "sum_a" - "mean_a" : non-numeric argument to binary operator 

Any ideas how to solve the problem!!!

Answer 1

You can use this function -

my_s_mutate <- function(df, group_var, expr) {
  expr1 <- enquo(expr)
  mean_name <- paste0("mean_", quo_name(expr1))
  sum_name <- paste0("sum_", quo_name(expr1))
  
  df %>%
    group_by({{group_var}}) %>%
    mutate(
      !! mean_name := mean({{expr}}),
      !! sum_name := sum({{expr}}),
      diff = .data[[sum_name]] - .data[[mean_name]]
    )
}

To work with variable names use {{}} , use strings as new column names with !! and := . Refer to column names as string with .data .

my_s_mutate(df, g1, a)

# A tibble: 5 x 7
# Groups:   g1 [2]
#     g1    g2     a     b mean_a sum_a  diff
#  <dbl> <dbl> <int> <int>  <dbl> <int> <dbl>
#1     1     1     1     2      3     6     3
#2     1     2     5     3      3     6     3
#3     2     1     2     4      3     9     6
#4     2     2     4     5      3     9     6
#5     2     1     3     1      3     9     6