I know it can be done more efficiently, but this appears to do what you are hoping for (hopefully in an understandable way). It differs that it uses the DATA handle instead of a separate file, and prints a space if there was no data for a particular name/set pair.
use strict;
use warnings;
my %hash;
my %found;
while (my $line = <DATA>) {
chomp $line;
next if ($line =~ m/^\s*$/);
my @part = split /\s+/, $line;
$found{$part[1]}++;
$hash{$part[0]}{$part[1]} = $part[2];
}
my @sets = sort keys %found;
print join(q{ }, q{NAME}, @sets), qq{\n};
foreach my $k (sort keys %hash) {
my @data;
foreach my $l (@sets) {
if (defined $hash{$k}{$l}) {
push @data, $hash{$k}{$l};
} else {
push @data, q{ };
}
}
print join(q{ }, $k, @data), qq{\n};
}
__DATA__
AA SET1 0
AA SET2 1
AA SET3 0
BB SET1 0
BB SET3 0
CC SET1 2
CC SET2 3
And the output:
NAME SET1 SET2 SET3
AA 0 1 0
BB 0 0
CC 2 3