How to use R data.table column names with cube(..., j = ,...) within a function?

Question

I can obtain a summary of a variable stratified by other variables as follows:

require(data.table)

DT <- data.table(mtcars)

var_work <- "hp"
by_vars <- c("cyl", "carb")

ans_1 <- cube(DT, j = as.list(quantile(get(var_work))), by = by_vars)

ans_1
    cyl carb  0%    25%   50%    75% 100%
 1:   6    4 110 110.00 116.5 123.00  123
 2:   4    1  65  66.00  66.0  93.00   97
 3:   6    1 105 106.25 107.5 108.75  110
 4:   8    2 150 150.00 162.5 175.00  175
 5:   8    4 205 218.75 237.5 245.00  264
 6:   4    2  52  69.25  93.0 105.50  113
 7:   8    3 180 180.00 180.0 180.00  180
 8:   6    6 175 175.00 175.0 175.00  175
 9:   8    8 335 335.00 335.0 335.00  335
10:   6   NA 105 110.00 110.0 123.00  175
11:   4   NA  52  65.50  91.0  96.00  113
12:   8   NA 150 176.25 192.5 241.25  335
13:  NA    4 110 123.00 210.0 241.25  264
14:  NA    1  65  66.00  93.0 101.00  110
15:  NA    2  52  92.00 111.0 150.00  175
16:  NA    3 180 180.00 180.0 180.00  180
17:  NA    6 175 175.00 175.0 175.00  175
18:  NA    8 335 335.00 335.0 335.00  335
19:  NA   NA  52  96.50 123.0 180.00  335

Next, I would like to write a helper function implementing exactly what is shown above, which however produces the error:

my_fun <- function(table_work, var_w, by_v) {

    tab_out <- cube(table_work, j = as.list(quantile(get(var_w))), by = by_v)
    return(tab_out)

}

ans_2 <- my_fun(table_work = DT, var_w = var_work, by_v = by_vars)

Error in get(var_w) : object 'var_w' not found

I have searched for the answer some relevant blogs (e.g., Advanced tips and tricks with data.table) and posts (e.g., by Henrik, frankc etc.), and tried different combinations of quote(), eval(), get(), assign() etc. within "my_fun", but nothing worked for me.

The question is: How should I correct "my_fun" helper function so it works and produces the same result as ans_1?

R version 3.6.1 (2019-07-05)
Platform: x86_64-apple-darwin15.6.0 (64-bit)
Running under: macOS Mojave 10.14.5

Matrix products: default
BLAS:   /Library/Frameworks/R.framework/Versions/3.6/Resources/lib/libRblas.0.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/3.6/Resources/lib/libRlapack.dylib

locale:
[1] en_AU.UTF-8/en_AU.UTF-8/en_AU.UTF-8/C/en_AU.UTF-8/en_AU.UTF-8

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] Matrix_1.2-18     fst_0.9.0         data.table_1.12.8

loaded via a namespace (and not attached):
[1] compiler_3.6.1  parallel_3.6.1  Rcpp_1.0.3      grid_3.6.1      lattice_0.20-38

chinsoon12 · Accepted Answer · 2020-01-07 06:05:33Z

4

When reading through the code for data.table:::cube.data.table and data.table:::groupingsets.data.table, the j argument is already being evaluated using NSE. Hence, being unable to pass in as.name(var_work) to the environment argument of substitute, the function will fail.

As a workaround, you can use .SDcols:

library(data.table)    
DT <- data.table(mtcars)    
var_work <- "hp"
by_vars <- c("cyl", "carb")

my_fun <- function(table_work, var_w, by_v) {
    cube(table_work, j=as.list(quantile(.SD[[1L]])), by=by_v, .SDcols=var_w)
}

ans_2 <- my_fun(table_work = DT, var_w = var_work, by_v = by_vars)

answered Jan 7, 2020 at 6:05

chinsoon12

25.2k4 gold badges27 silver badges35 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

chinsoon12 Over a year ago

@jangorecki is the best person to answer this question imo. if he comes along, i will be glad to delete this answer

Max Moldovan Over a year ago

it works, thanks. Still, why cube(), and I guess other functions, cannot see 'var_w' even though it is explicitly declared within "my_fun"?

chinsoon12 Over a year ago

var_w is visible in the scope of my_fun. however, it is passed as jj = substitute(j) into cube.data.table and groupingsets.data.table, and jj is eval in the scope of the groupingsets.data.table which does not have var_w.

jangorecki Over a year ago

Just came across this Q, pinging my username didn't work AFAICT. This is good workaround. Of course we are open for suggestions or improvement if needed.

Roland · Accepted Answer · 2020-01-07 06:50:55Z

3

You can always compute on the language:

my_fun <- function(table_work, var_w, by_v) {

  var_w <- as.name(var_w)
  eval(bquote(cube(table_work, j = as.list(quantile(.(var_w))), by = by_v)))  

}

answered Jan 7, 2020 at 6:50

Roland

134k12 gold badges203 silver badges305 bronze badges

Collectives™ on Stack Overflow

How to use R data.table column names with cube(..., j = ,...) within a function?

2 Answers 2

4 Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

4 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related