view release on metacpan or  search on metacpan

lib/AI/Categorizer/Learner/Weka.pm  view on Meta::CPAN

  foreach my $name (@$feature_names) {
    print $fh "\@ATTRIBUTE feature-$name REAL\n";
  }
  print $fh "\@ATTRIBUTE category {1, 0}\n\n";
  
  my %feature_indices = map {$feature_names->[$_], $_} 0..$#{$feature_names};
  my $last_index = keys %feature_indices;
  
  # We use the 'sparse' format, see http://www.cs.waikato.ac.nz/~ml/weka/arff.html
  
  print $fh "\@DATA\n";
  foreach my $doc (@$docs) {
    my ($features, $cat) = @$doc;
    my $f = $features->as_hash;
    my @ordered_keys = (sort {$feature_indices{$a} <=> $feature_indices{$b}} 
			grep {exists $feature_indices{$_}}
			keys %$f);

    print $fh ("{",
	       join(', ', map("$feature_indices{$_} $f->{$_}", @ordered_keys), "$last_index '$cat'"),
	       "}\n"