Getting error while case_when() applied on empty column in R-CodePudding

I have several dataframes with same column names. I wrote a code to perform some operations on a column based on several conditions. But I got error if all the values of that column is empty.

For an example, if my dataframe is like below, my code works fine.

unit	temperature
C	70
C
F	20
K	130

But if my dataframe is like below, my code gives error.

unit	temperature
K
C
F
C

My code is,

data <- data %>%
  mutate(temperature = case_when(
               unit == 'C' ~ temperature, 
               unit == 'F' ~ (temperature - 32)*(5/9),
               unit == 'K' ~ temperature - 273))

Could someone help?

CodePudding user response：

You can force your empty column (it does not matter if it contains empty strings or logical NA's) to a specific data type, i.e. in this case to a numeric class. Then your case_when() should work and should return NA (NA_real_)

library(dplyr)    

tibble::tribble(~unit, ~temperature,
                    "K","", 
                    "C","", 
                    "F","", 
                    "C", "") %>% 
      mutate(
        temperature = temperature %>% as.numeric(),
        temperature = case_when(
        unit == 'C' ~ temperature, 
        unit == 'F' ~ (temperature - 32)*(5/9),
        unit == 'K' ~ temperature - 273))

CodePudding user response：

library(tidyverse)

data <- tribble(
  ~unit, ~temperature,
  "K", NA,
  "C", NA,
  "F", NA,
  "C", NA
)

data %>%
  mutate(
    temperature = as.numeric(temperature),
    temperature = case_when(
      unit == "C" ~ temperature,
      unit == "F" ~ (temperature - 32) * (5 / 9),
      unit == "K" ~ temperature - 273
    )
  )
#> # A tibble: 4 × 2
#>   unit  temperature
#>   <chr>       <dbl>
#> 1 K              NA
#> 2 C              NA
#> 3 F              NA
#> 4 C              NA

^{Created on 2022-06-08 by the reprex package (v2.0.0)}