From: Steinar H. Gunderson <sgunderson@bigfoot.com>
Date: Wed, 30 May 2012 19:06:44 +0000 (+0200)
Subject: Split train.pl into functions.
X-Git-Url: https://git.sesse.net/?p=wloh;a=commitdiff_plain;h=df74bf90cbb758f689a4e0d5f19f77d74beabd8b

Split train.pl into functions.
---

diff --git a/train.pl b/train.pl
index d128d58..daa4d78 100755
--- a/train.pl
+++ b/train.pl
@@ -6,19 +6,16 @@ no warnings qw(once);
 use POSIX;
 require './config.pm';
 
-my $dbh = DBI->connect($config::local_connstr, $config::local_username, $config::local_password)
-	or die "connect: " . $DBI::errstr;
-$dbh->{AutoCommit} = 0;
-$dbh->{RaiseError} = 1;
-
-$dbh->do('SET client_min_messages TO WARNING');
-
 # Find last completely done season 
-my $ref = $dbh->selectrow_hashref('SELECT sesong FROM fotballserier GROUP BY sesong HAVING COUNT(*)=COUNT(avgjort=1 OR NULL) ORDER BY sesong DESC LIMIT 1');
-my $last_season = $ref->{'sesong'};
+sub find_last_season {
+	my $dbh = shift;
+	my $ref = $dbh->selectrow_hashref('SELECT sesong FROM fotballserier GROUP BY sesong HAVING COUNT(*)=COUNT(avgjort=1 OR NULL) ORDER BY sesong DESC LIMIT 1');
+	return $ref->{'sesong'};
+}
 
-# Fetch games
-my $q = $dbh->prepare('
+sub fetch_games {
+	my ($dbh, $last_season, $games, $ids) = @_;
+	my $q = $dbh->prepare('
 SELECT
   deltager1.id as p1, deltager2.id as p2, maalfor, maalmot, least(pow(2.0, (sesong - ? + 3) / 3.0), 1.0) AS vekt
 FROM
@@ -27,51 +24,75 @@ FROM
   JOIN Fotballdeltagere deltager2 ON resultater.Motstander=deltager2.Nr AND resultater.Serie=deltager2.Serie
   JOIN Fotballserier serier on resultater.Serie=serier.Nr
 WHERE deltager1.Nr > deltager2.nr
-');
-$q->execute($last_season);
-
-my @games = ();
-my %ids = ();
+	');
+	$q->execute($last_season);
 
-while (my $ref = $q->fetchrow_hashref) {
-	next if ($ref->{'maalfor'} == 150 && $ref->{'maalmot'} == 0);
-	next if ($ref->{'maalfor'} == 0 && $ref->{'maalmot'} == 150);
-	next if ($ref->{'maalfor'} == 150 && $ref->{'maalmot'} == 150);
-	next if ($ref->{'maalfor'} == 0 && $ref->{'maalmot'} == 0);
-	push @games, { %$ref };
-	$ids{$ref->{'p1'}} = 1;
-	$ids{$ref->{'p2'}} = 1;
+	while (my $ref = $q->fetchrow_hashref) {
+		next if ($ref->{'maalfor'} == 150 && $ref->{'maalmot'} == 0);
+		next if ($ref->{'maalfor'} == 0 && $ref->{'maalmot'} == 150);
+		next if ($ref->{'maalfor'} == 150 && $ref->{'maalmot'} == 150);
+		next if ($ref->{'maalfor'} == 0 && $ref->{'maalmot'} == 0);
+		push @$games, { %$ref };
+		$ids->{$ref->{'p1'}} = 1;
+		$ids->{$ref->{'p2'}} = 1;
+	}
 }
 
-# Output to file
-my $tmpnam = POSIX::tmpnam();
-open DATA, ">", $tmpnam
-	or die "$tmpnam: $!";
+sub output_to_file {
+	my ($games, $ids) = @_;
 
-printf DATA "%d\n", scalar keys %ids;
-for my $id (keys %ids) {
-	printf DATA "%d\n", $id;
-}
-for my $ref (@games) {
-	printf DATA "%d %d %d %d %f\n", $ref->{'p1'}, $ref->{'p2'}, $ref->{'maalfor'}, $ref->{'maalmot'}, $ref->{'vekt'};
+	my $tmpnam = POSIX::tmpnam();
+	open DATA, ">", $tmpnam
+		or die "$tmpnam: $!";
+
+	printf DATA "%d\n", scalar keys %$ids;
+	for my $id (keys %$ids) {
+		printf DATA "%d\n", $id;
+	}
+	for my $ref (@$games) {
+		printf DATA "%d %d %d %d %f\n", $ref->{'p1'}, $ref->{'p2'}, $ref->{'maalfor'}, $ref->{'maalmot'}, $ref->{'vekt'};
+	}
+	close DATA;
+
+	return $tmpnam;
 }
-close DATA;
 
-my @ratings = ();
-my @covariances = ();
+sub train_model {
+	my ($filename, $ratings, $covariances) = @_;
 
-open RATINGS, "$config::base_dir/bayeswf < $tmpnam |"
-	or die "bayeswf: $!";
-while (<RATINGS>) {
-	chomp;
-	my @x = split;
-	if ($x[0] eq 'covariance') {
-		push @covariances, (join("\t", @x[1..3]));
-	} else {
-		push @ratings, ($x[2] . "\t" . $x[0] . "\t" . $x[1]);
+	open RATINGS, "$config::base_dir/bayeswf < $filename |"
+		or die "bayeswf: $!";
+	while (<RATINGS>) {
+		chomp;
+		my @x = split;
+		if ($x[0] eq 'covariance') {
+			push @$covariances, (join("\t", @x[1..3]));
+		} else {
+			push @$ratings, ($x[2] . "\t" . $x[0] . "\t" . $x[1]);
+		}
 	}
+
+	close RATINGS;
 }
 
+my $dbh = DBI->connect($config::local_connstr, $config::local_username, $config::local_password)
+	or die "connect: " . $DBI::errstr;
+$dbh->{AutoCommit} = 0;
+$dbh->{RaiseError} = 1;
+
+$dbh->do('SET client_min_messages TO WARNING');
+
+my $last_season = find_last_season($dbh);
+my @games = ();
+my %ids = ();
+fetch_games($dbh, $last_season, \@games, \%ids);
+my $tmpnam = output_to_file(\@games, \%ids);
+
+my @ratings = ();
+my @covariances = ();
+train_model($tmpnam, \@ratings, \@covariances);
+unlink($tmpnam);
+
 $dbh->do('TRUNCATE ratings');
 $dbh->do('COPY ratings ( id, rating, rating_stddev ) FROM STDIN');
 $dbh->pg_putcopydata(join("\n", @ratings));
@@ -86,4 +107,3 @@ $dbh->do('DROP TABLE IF EXISTS covariance');
 $dbh->do('ALTER TABLE new_covariance RENAME TO covariance');
 
 $dbh->commit;
-unlink($tmpnam);