Gnuplot, CSV, only distinct labels?

Question

I need some help on processing a CSV file with gnuplot. My CSV file looks like:

a,1002
b,612
b,893
b,361
b,932
b,483
b,899
b,614
c,722
d,1038
d,1580

I want it to use the 2nd column for Y value, the line number as X value. This is done easily.

I managed to put the first column as xtics label, but after hours of Google-ing, I did not find a solution to tell gnuplot to only print a label only once (my x axis was crowded with labels, making it unreadable).

I need something like:

.
.     b
.a ***** c
.**    ***       d  ****
.        ************
........................

But I'm getting this:

    .
    .  
    .  *****
    .**     ***          ****
    .        ************
    ........................
   sdflksdjflksdjflksdjflkadjflksjdflkasj

Labels will be okay on the axis or on the graph line too, it doesn't matter.

Answer 1

To set xtics that are not periodic, use

set xtics ("a" 1, "b" 2, "c" 9, "d" 10)

You can extract the information from your data by a simple script, e.g.

cut -f1 -d, data | while read name ; do
    let line++
    if [[ $name != $oldname ]] ; then
        oldname=$name
        echo "'$name'" $line
    fi
done

Answer 2

Some further suggestions:

set pointsize 2.5
set xtics ("a" 1, "b" 5, "c" 9, "d" 10.5)
set xrange [0:12]
plot "< cat -n aa.csv | sed -e 's/,/  /g'" using 1:3 pt 7 notitle

yields unequally spaced labels in place of x-ticks (as suggested by @choroba, which also provided a way to automate the determination of mid-interval for larger datasets). (Here aa.csv contains data as shown in your question.)

If you want labels to appear in the plotting region rather than under the x-axis, then you can start with

set format x ""

and then place your labels above the highest value found in any of the above categories, a, ..., d.