\uxxxx sequences inside backticks - r

Question

I know that, in general, \uxxxx sequences are not supported inside backticks. Do you have any workaround to include them (\uxxxx sequences) in column names?
To be specific, imagine calculating Body Mass Index and adding units to column names.
Start with
dt<-data.frame(
`Weight [kg]` = runif(5,50,100),
`Height [m]` = runif(5,1.5,2),
check.names=F
)
and mutate:
> dt2<-dt %>%
mutate(
`BMI [kg/m\u00b2]`= `Weight [kg]`/`Height [m]`^2
)
This produces an error: Error: \uxxxx sequences not supported inside backticks (line 3).
My workaround is like this:
> dt2<-dt %>%
mutate(
`BMI [kg/m2]`= `Weight [kg]`/`Height [m]`^2
) %>%
set_colnames(colnames(.) %>% str_replace('2\\]', '\u00b2\\]'))
> colnames(dt2)
[1] "Weight [kg]" "Height [m]" "BMI [kg/m²]"
It gives me exactly what I want but is not very elegant.
Suprisingly, a bit clearer approach fails:
> dt2<-dt %>%
mutate(
`BMI [kg/m2]`= `Weight [kg]`/`Height [m]`^2
) %>%
rename_all(str_replace, '2\\]', '\u00b2\\]')
> colnames(dt2)
[1] "Weight [kg]" "Height [m]" "BMI [kg/m2]"
So, my question is: can it be done in not-so-hacky way?
And:
yes, I'm sure, I need \uxxx in column names;
yes, I use them further on graphs;
no, I don't want to replace them with expression's.

Answer 1

How about just using single quotes instead of backticks?
dt %>% mutate('BMI [kg/m\u00b2]' = `Weight [kg]`/`Height [m]`^2)
#> Weight [kg] Height [m] BMI [kg/m²]
#> 1 67.68154 1.757490 21.91211
#> 2 72.32362 1.817616 21.89151
#> 3 89.28197 1.854459 25.96146
#> 4 52.14819 1.709520 17.84395
#> 5 83.48281 1.969367 21.52502
Or double quotes?
dt %>% mutate("BMI [kg/m\u00b2]" = `Weight [kg]`/`Height [m]`^2)
#> Weight [kg] Height [m] BMI [kg/m²]
#> 1 67.68154 1.757490 21.91211
#> 2 72.32362 1.817616 21.89151
#> 3 89.28197 1.854459 25.96146
#> 4 52.14819 1.709520 17.84395
#> 5 83.48281 1.969367 21.52502
You can also use them to access items in your new data frame:
dt2$'BMI [kg/m\u00b2]'
#> [1] 21.91211 21.89151 25.96146 17.84395 21.52502
dt2$"BMI [kg/m\u00b2]"
#> [1] 21.91211 21.89151 25.96146 17.84395 21.52502
Or did you specifically need to use backticks for some reason?

Answer 2

Argument names don't have to be in backticks, they can be regular quoted strings. So this works fine:
dt2<-dt %>%
mutate(
"BMI [kg/m\u00b2]" = `Weight [kg]`/`Height [m]`^2
)
It will be hard to refer to that column name in expressions in later code; you'll need to specify the column by number, or use an expression like dt2["BMI [kg/m\u00b2]"] (or dt2$"BMI [kg/m\u00b2]" as used by #AllanCameron in his answer). But it will print fine:
> dt2
Weight [kg] Height [m] BMI [kg/m²]
1 51.89918 1.825124 15.58029
2 80.74140 1.602126 31.45595
3 71.35380 1.974187 18.30799
4 64.44167 1.989202 16.28580
5 76.13564 1.886232 21.39922
Edited to add: It's also fine to use
`BMI [kg/m²]`
anywhere a column name can be used, you just can't encode the special char with \uxxxx.