Compute Differences of Two Columns

The default delimiter in awk is white space, but csv uses quotes and commas, so awk by default will see the csv file as one big column.

Also, you'll need to strip the quotes for it to recognize the numbers. So you can use -F'","' to change the delimiter to quoted strigns separated by comma, which will work as long as none of your quoted strings contain just a comma.

awk -F'","'  '{ print $2 - $3 }' filename.csv

Note that you have to quote the quotes for the shell.

This works:

$ awk 'BEGIN{FS=OFS="\""}{$7=",\""$4-$6"\""}1' filename.csv

Output:

"Afghanistan","94.0","81.1","12.9"
"Bahamas","42.9","43.2","-0.3"
"Bolivia (Plurinational State of)","86.7","31.9","54.8"
"Brazil","76.7","0.0","76.7"

As an aside, IF the number fields were unquoted, it would be a lot simpler: awk -F, '{$4=$2-$3}1' filename.csv.

Using Raku (formerly known as Perl_6)

raku -MText::CSV -e 'my @a = csv(in => "/Path/To/File");  \
      my @b = @a.map(*.[1]) Z- @a.map(*.[2]); \ 
     .put for @a Z @b;'

#OR

raku -MText::CSV -e 'my @a = csv(in => "/Path/To/File");  \
      .put for @a Z (@a.map(*.[1]) Z- @a.map(*.[2]));' 

Sample Input:

"Afghanistan","94.0","81.1"
"Bahamas","42.9","43.2"
"Bolivia (Plurinational State of)","86.7","31.9"
"Brazil","76.7","0.0"

Sample Output:

Afghanistan 94.0 81.1 12.9
Bahamas 42.9 43.2 -0.3
Bolivia (Plurinational State of) 86.7 31.9 54.8
Brazil 76.7 0.0 76.7

While I'm sure you can 'get-away-with' parsing csv on your own (for small projects), it may make sense at some point for you to use a CSV parser (to handle embedded quotes, embedded newlines, detect/change comma/non-comma column separators, trim whitespace, etc.).

You might be surprised how easy it is to invoke a CSV parser at the bash command line in the Raku programming language. First load theText::CSV module with the -M (module) commandline flag, which gives you -MText::CSV. Then tell Raku to expect code to follow with the -e commandline flag.

To explain the first example, an @a array is declared with my and takes csv input. A @b array is declared with my that takes the difference between (zero-indexed) column 1 and column 2. This difference is accomplished with the Z zip metaoperator appended with the minus operator, to make Z-. Finally, the Z zip metaoperator is used to output the original @a csv, zipped row-by-row with the @b column.

https://github.com/Tux/CSV/blob/master/doc/Text-CSV.pdf (click the Download link)
https://docs.raku.org/language/operators#index-entry-Z_(zip_metaoperator)
https://raku.org

Using Miller (mlr) to read the header-less CSV file, create a new fourth field by computing the difference between the 2nd and 3rd field, format the resulting floating-point value according to the standard %g format specifier, and output quoted CSV:

$ mlr --csv -N --quote-all put '$4 = fmtnum($2 - $3, "%g")' file
"Afghanistan","94.0","81.1","12.9"
"Bahamas","42.9","43.2","-0.3"
"Bolivia (Plurinational State of)","86.7","31.9","54.8"
"Brazil","76.7","0.0","76.7"

Note that the %g formatting seems to be broken in later releases of Miller, so you may need to use something like %.6g instead.