gnuplot timeseries with categorical states as step function

I have a datafile which looks something like this:

0;State a
1;State a
2;State b
3:State b
4:State a

Where the first column represents the time in seconds, and the second column represents a certain state.

I want to plot the occurences of the events in gnuplot over time. I am trying to use the following for plotting:

set datafile separator ";"
plot 'data' using 1:2:yticlabels(2)

However I get the following error:

warning: Skipping data file with no valid points
                                       ^
x range is invalid

It seems like gnuplot won't recognize the strings as categorical values. The result should look something like a non-continous step function:

       ^
State b┼       ┌───────┐
       │       │       │
State a┼───────┘       └────
       │
       ┼───┼───┼───┼───┼───┼─>
       0   1   2   3   4   5           

Is this sort of plot possible with gnuplot? If, so how would you do this?

No, gnuplot doesn't recognize strings as categorical values. You must do those assignments "string → integer" yourself.

The easiest way to do this mapping is to use an external tool like awk and add the integer values on-the-fly. The following awk call does this mapping and adds the values to the output:

awk -F ';' -v OFS=';' '{
  if (!($2 in array)) { 
    array[$2] = length(array)
  }
  print $1,$2,array[$2]
}' data.csv

Using the gnuplot syntax

plot "< awk ..."

you can combine the awk call directly with the plotting:

set datafile separator ";"
set offset 0.1,0.1,0.1,0.1
set xtics 0,1
plot "< awk -F ';' -v OFS=';' '{if (!($2 in array)) { array[$2] = length(array) }; print $1,$2,array[$2]}' data.csv" using 1:3:ytic(2) w step lw 3 notitle

The output is

Alternatively, if you haven't access to awk, you can do the preprocessing also using e.g. a python script like the following cat.py:

from __future__ import print_function
import sys
a={}
with open(sys.argv[1], 'r') as f:
    for line in f:
        fields = line.strip().split(';')
        if (not fields[1] in a):
            a[fields[1]] = len(a)
        print("{0};{1};{2}".format(fields[0], fields[1], a[fields[1]]), file=sys.stdout)

and call it with

plot "< python cat.py data.csv" ...

Side note: maybe one could also do this using gnuplot only, but that can become quite ugly, see Gnuplot, plotting a graph with text on y axis for a similar use case.