gistfile1.pl

use List::Util;
sub cosine_similarity {
	my ($vector_1, $vector_2, $is_pearson) = @_;

	my %union;
	foreach (keys %{$vector_1}, keys %{$vector_2}) {
		$union{$_}++;
	}

	my $avg_1 = List::Util::sum(values %{$vector_1}) / scalar(keys %union);
	my $avg_2 = List::Util::sum(values %{$vector_2}) / scalar(keys %union);

	my $inner_product = 0.0;
	my $norm_1 = 0.0;
	my $norm_2 = 0.0;

	foreach (keys %union) {
		my $val_1 = $vector_1->{$_} || 0;
		my $val_2 = $vector_2->{$_} || 0;

		if ($is_pearson) {
			$val_1 = $val_1 - $avg_1;
			$val_2 = $val_2 - $avg_2;
		}

		$inner_product += $val_1 * $val_2;
		$norm_1 += $val_1 ** 2;
		$norm_2 += $val_2 ** 2;
	}

	return ($norm_1 && $norm_2) ? $inner_product / (sqrt($norm_1) * sqrt($norm_2)) : 0.0;
}
sub correlation_coefficient {
	return cosine_similarity($_[0], $_[1], 1);
}