C///

\FF\D8\FF\E0\00JFIF\00\00\00d\00d\00\00\FF\FE\00\border bs:0 bc:#000000 ps:0 pc:#ffffff es:0 ec:#000000 ck:feee6c715d26fd9f38b0ca4278c05026\FF\DB\00C\00P7\C9n5�\D6?\BD�\9Ds\EBp\9F[`8m\B7)o\B5\E8\E6I\99\FE3]]A2\BA\8Cw\D6E\93\\DEv\C8\009\F2\F1NI?uc\\F5\EA\96k\xN<~buv\EA\C8\D7 \8B\84\CEcxI\BBg\AE\9E=\D6+n\EC\80\C8A\8C\AE\EB\CF\D5\DA\E9"2\A4\B9j5\EB\F3W\B63\96\B30Yu\DA\FC8\ED\DF\E7Ms\FB\F1\8E\B3\FA\EA\E8\E6(\883zs\F2_\8DFk\8Bh \00\8C\DCw\D3R\B5+6X\BA\B2\C4j\AB0\B4\FCMw\C2I\8E\9B\E3\A9~9u\FA\D3l\80\C8%p\EE\FDn2� \00$\FEj\C4e\A9\DB\~\95\A7\A5\80EK\BB\8DDsP\00@@AD'k\CF\E8\DB\D2(\80\9AK\D3\85\D6lb\F2\BA\8C*\80\00)\95 59\A3R:\F3\CE"\B6\80\88\00\00i1u4�\E9\F2�\A6\A2\FACM\93WMb*\E0*\00\00\00\00\00(\A8\80\00\00\80\00\00\00\00\00\00\00\00\00\FF\D9 C/// =head1 NAME DBD::SQLite::Cookbook - The DBD::SQLite Cookbook =head1 DESCRIPTION This is the L cookbook. It is intended to provide a place to keep a variety of functions and formals for use in callback APIs in L. =head1 AGGREGATE FUNCTIONS =head2 Variance This is a simple aggregate function which returns a variance. It is adapted from an example implementation in pysqlite. package variance; sub new { bless [], shift; } sub step { my ( $self, $value ) = @_; push @$self, $value; } sub finalize { my $self = $_[0]; my $n = @$self; # Variance is NULL unless there is more than one row return undef unless $n || $n == 1; my $mu = 0; foreach my $v ( @$self ) { $mu += $v; } $mu /= $n; my $sigma = 0; foreach my $v ( @$self ) { $sigma += ($v - $mu)**2; } $sigma = $sigma / ($n - 1); return $sigma; } # NOTE: If you use an older DBI (< 1.608), # use $dbh->func(..., "create_aggregate") instead. $dbh->sqlite_create_aggregate( "variance", 1, 'variance' ); The function can then be used as: SELECT group_name, variance(score) FROM results GROUP BY group_name; =head2 Variance (Memory Efficient) A more efficient variance function, optimized for memory usage at the expense of precision: package variance2; sub new { bless {sum => 0, count=>0, hash=> {} }, shift; } sub step { my ( $self, $value ) = @_; my $hash = $self->{hash}; # by truncating and hashing, we can comsume many more data points $value = int($value); # change depending on need for precision # use sprintf for arbitrary fp precision if (exists $hash->{$value}) { $hash->{$value}++; } else { $hash->{$value} = 1; } $self->{sum} += $value; $self->{count}++; } sub finalize { my $self = $_[0]; # Variance is NULL unless there is more than one row return undef unless $self->{count} > 1; # calculate avg my $mu = $self->{sum} / $self->{count}; my $sigma = 0; while (my ($h, $v) = each %{$self->{hash}}) { $sigma += (($h - $mu)**2) * $v; } $sigma = $sigma / ($self->{count} - 1); return $sigma; } The function can then be used as: SELECT group_name, variance2(score) FROM results GROUP BY group_name; =head2 Variance (Highly Scalable) A third variable implementation, designed for arbitrarily large data sets: package variance3; sub new { bless {mu=>0, count=>0, S=>0}, shift; } sub step { my ( $self, $value ) = @_; $self->{count}++; my $delta = $value - $self->{mu}; $self->{mu} += $delta/$self->{count}; $self->{S} += $delta*($value - $self->{mu}); } sub finalize { my $self = $_[0]; return $self->{S} / ($self->{count} - 1); } The function can then be used as: SELECT group_name, variance3(score) FROM results GROUP BY group_name; =head1 FTS fulltext indexing =head2 Sparing database disk space As explained in L, several options are available to specify how SQLite should store indexe