Gnuplot - plotting csv file based on value on row

2
votes

I am new to gnuplot. I have csv file like this:

category| date      | value
 a      |2016-04-01 | 0,2
 a      |2016-04-02 | 0,31
 a      |2016-04-03 | 0,14
....
 a      |2016-04-11 | 0,4
 b      |2016-04-01 | 0,32
 b      |2016-04-02 | 0,31
....
 b      |2016-04-10 | 0,15
 C      |2016-04-01 | 0,15
 C      |2016-04-02 | 0,23
.....

And so on

How do I plot data for each category on different graph in one command?

For example, if I have 9 distinct categories, then I have 9 graphs where x axis is for date, and y axis is for value.

2
rcsvgnuplot
why is the R tag there? - hrbrmstr
This is not a csv file. csv is the abbreviation for "Comma Separated Values". And if you have any influence on the format: switch the decimal separator to ".", and separate columns by space or tab. - Karl

2 Answers

1
votes

You can plot the file several times using plots for-loop. Each time, you only regard x values where column one fits the loop variable (check help ternary operator):

set decimalsign ","
set xdata time
set format x "%Y-%M-%D"
plot for [cat in "a b c d"] datafile using ($1 eq cat ? $2 : NaN):3

Gnuplot cannot connect the points (e.g. with linespoints) if there are invalid points in between. So you'll have to sort the file beforhand.

0
votes

I used Perl to reformat the input for gnuplot.

#! /bin/bash
input=$1
tmp=$(mktemp)
perl -aF'/\|/' \
    -ne 'print "\n\n" if $last ne $F[0] && $. > 2;
         $last = $F[0];
         $F[-1] =~ s/,/./;
         print "@F" if $. > 1;
        ' "$input" > "$tmp"

categories=($( cut -f1 -d\| "$input" | uniq ))

{
    cat <<EOF
    set term png;
    set xdata time;
    set timefmt '%Y-%m-%d';
    plot '$tmp' index 0 using 2:3 with lines title "${categories[1]}" \\
EOF

    i=1
    while (( i < ${#categories[@]}-1 )) ; do
        printf ', "" index %d using 2:3 with lines title "%s"' $i "${categories[++i]}"
    done
} | gnuplot > "$input".png

rm "$tmp"