gnuplot: How to put sum above bars with clustered stacked histograms

Question

Thanks to the question gnuplot rowstacked histogram: how to put sum above bars and its answer, I learned how to add the sum of stacked histograms above their bars. However, I can't figure out how to achieve this with stacked histograms that are clustered. As shown in the below code, I can get the sum which seem to be placed at the right y coordinates.

Unfortunately, I don't know how to change the x coordinates for the second and third clusters, so they end up overlapping the first cluster's sum labels. What do I need to do to change the x coordinates of the sums so that they appear over the correct bars?

set term png size 1200, 500
set output "graph.png"
set datafile separator ","

set boxwidth 0.5
set style fill solid
set style data histograms
set style histogram rowstack

set offset -1.3,-0.3,100,0
set yrange [*:*] noextend
set xrange [*:*] noextend

set key outside
set key right top
set ylabel "Time (ms)"
set xtics font ", 10"

plot newhistogram at 0, "a.csv" u 2:xtic(1) t "Download" lc 1, \
    '' u 3:xtic(1) t "Transfer" lc 2, \
    '' u 4:xtic(1) t "Return" lc 3, \
    '' u 5:xtic(1) t "Processing" lc 4, \
    '' u 6:xtic(1) t "Wait" lc 5, \
    '' u 7:xtic(1) t "Overhead" lc 6, \
    '' u 0:($2+$3+$4+$5+$6+$7):(sprintf('%d', $2+$3+$4+$5+$6+$7)) not with labels offset 0,0.7, \
  newhistogram at 3, "b.csv" u 2:xtic(1) t "" lc 1,\
    for [i=3:7] '' u i:xtic(1) t "" lc (i - 1), \
    '' u 0:($2+$3+$4+$5+$6+$7):(sprintf('%d', $2+$3+$4+$5+$6+$7)) not with labels offset 0,0.7, \
  newhistogram at 6, "c.csv" u 2:xtic(1) t "" lc 1,\
    for [i=3:7] '' u i:xtic(1) t "" lc (i - 1), \
    '' u 0:($2+$3+$4+$5+$6+$7):(sprintf('%d', $2+$3+$4+$5+$6+$7)) not with labels offset 0,0.7

Data files:

• a.csv

  Device,Download,Transfer,Return,Processing,Wait,Overhead
  Find X2 Pro,350,0,0,287,1,24
  OnePlus 8,350,29,19,410,30,138
  

• b.csv

  Device,Download,Transfer,Return,Processing,Wait,Overhead
  Find X2 Pro,350,0,0,293,1,26
  Pixel 6,350,81,21,216,100,228
  

• c.csv

  Device,Download,Transfer,Return,Processing,Wait,Overhead
  Pixel 6,350,0,0,436,2,43
  Pixel 3,350,140,46,187,52,206
  

Answer 1

Here I tried to clean up and simplify your script. I included the data as datablocks, so it is simply copy & paste & run. Check help sum, help columnheader... I hope the rest is self-explaining.

Script:

### rowstacked histogram with total above bars
reset session

$DataA <<EOD
Device,Download,Transfer,Return,Processing,Wait,Overhead
Find X2 Pro,350,0,0,287,1,24
OnePlus 8,350,29,19,410,30,138
EOD

$DataB <<EOD
Device,Download,Transfer,Return,Processing,Wait,Overhead
Find X2 Pro,350,0,0,293,1,26
Pixel 6,350,81,21,216,100,228
EOD

$DataC <<EOD
Device,Download,Transfer,Return,Processing,Wait,Overhead
Pixel 6,350,0,0,436,2,43
Pixel 3,350,140,46,187,52,206
EOD

set datafile separator ","
set boxwidth 0.5
set style fill solid
set style data histograms
set style histogram rowstack

set offset -1.3,-0.3,100,0
set xrange [0:]
set xtics font ", 10" right rotate by 45
set yrange [0:*]
set ylabel "Time (ms)"
set key outside right top noautotitle

mySum(n) = sum [_i=2:7] column(_i)

plot newhistogram at 2, \
         for [i=2:7] $DataA u i:xtic(1) ti columnheader(i) lc i-1, \
         '' u ($0+1):(total=mySum(0)):(sprintf('%d', total)) w labels offset 0,0.7, \
     newhistogram at 4,\
         for [i=2:7] $DataB u i:xtic(1) lc i-1, \
         '' u ($0+4):(total=mySum(0)):(sprintf('%d', total)) w labels offset 0,0.7, \
     newhistogram at 7, \
         for [i=2:7] $DataC u i:xtic(1) lc i-1, \
         '' u ($0+7):(total=mySum(0)):(sprintf('%d', total)) w labels offset 0,0.7, \
### end of script

Result: