CODEWITHSUNDEEP
rdplyr
zhiwei li
zhiwei li

Reputation: 1711

Can not use Non-Standard Evaluation in dplyr in r

I want to rowsum some variables with the variables' name.

I tried some ways but all fail.

How can I run sum(ttt) correctly?

Any help will be highly appreciated!

library(dplyr) 
data(mtcars)
m2 = mtcars[1:3,10:11]

# Run
m2 %>% 
  rowwise() %>% 
  mutate(cc = sum(carb, gear))
# Not Run
m2 %>% 
  rowwise() %>% 
  mutate(cc = sum('carb', 'gear'))
# Not Run
ttt = c('carb', 'gear')
m2 %>% 
  rowwise() %>% 
  mutate(cc = sum(ttt))
# Run but the result was wrong
m2 %>% 
  rowwise() %>% 
  mutate(cc = sum(get(ttt)))
# Not Run
m2 %>% 
  rowwise() %>% 
  mutate(cc = sum(mget(ttt)))

Upvotes: 0

Views: 39

Answers (2)

akrun
akrun

Reputation: 886938

We can convert to symbols and evaluate (!!!)

library(dplyr)
m2 %>% 
    rowwise() %>% 
    mutate(cc = sum(!!! rlang::syms(c('carb', 'gear'))))

-output

# A tibble: 3 x 3
# Rowwise: 
#   gear  carb    cc
#  <dbl> <dbl> <dbl>
#1     4     4     8
#2     4     4     8
#3     4     1     5

Or instead use rowSums on the subset which would be vectorized and fast

m2 %>%
   mutate(cc = rowSums(select(., 'carb', 'gear')))

select can take both quoted and unquoted column names

Upvotes: 2

stefan
stefan

Reputation: 123783

You have to make use of dplyr::c_across:

While it works without all_of it throws a warning so using all_of is recommended if you a vector of characters.

library(dplyr) 
data(mtcars)
m2 = mtcars[1:3,10:11]

ttt = c('carb', 'gear')
m2 %>% 
  rowwise() %>% 
  mutate(cc = sum(c_across(all_of(ttt))))
#> # A tibble: 3 x 3
#> # Rowwise: 
#>    gear  carb    cc
#>   <dbl> <dbl> <dbl>
#> 1     4     4     8
#> 2     4     4     8
#> 3     4     1     5

Created on 2020-12-24 by the reprex package (v0.3.0)

Upvotes: 2

Related Questions