masses/bayes-testing/map-s-space/run-search

#!/usr/bin/perl -w

$USE_CHI_COMBINING = 0;

# use 1 bucket's set for speed, just to get going:
#$SPAMLOG  = "/raid/bayestest/results/bucket1/spam.log";
#$HAMLOG   = "/raid/bayestest/results/bucket1/nonspam.log";
# or use the lot, slow but more accurate:
$SPAMLOG  = "/raid/bayestest/results/spam_all.log";
$HAMLOG   = "/raid/bayestest/results/nonspam_all.log";

my $searchspaces = {
 ROBINSON_S_CONSTANT => [ 0.0, 0.75 ],
 ROBINSON_X => [ 0.0, 0.75 ],
};

my $count;
$|=1;
srand(23);
for ($count = 0; $count < 1000; $count++) {
  my $rands = rand_constants();
  write_constants($rands);
  run_test($count);
  parse_log($rands, $count);
}
exit;

sub rand_constants {
  my $rands = { };
  foreach (keys %{$searchspaces}) {
    my $lo = $searchspaces->{$_}->[0];
    my $hi = $searchspaces->{$_}->[1];

    $rands->{$_} = rand($hi-$lo) + $lo;
  }
  return $rands;
}

sub write_constants {
  my $rands = shift;
  open (OUT, ">constants.pl");
  print OUT "
    use constant ROBINSON_S_CONSTANT => $rands->{ROBINSON_S_CONSTANT};
    use constant ROBINSON_X => $rands->{ROBINSON_X};
    use constant ROBINSON_MIN_PROB_STRENGTH => 0.27;
    use constant PROB_BOUND_LOWER => 0.001;
    use constant PROB_BOUND_UPPER => 0.999;
    use constant N_SIGNIFICANT_TOKENS => 150;
    use constant USE_CHI_COMBINING => $USE_CHI_COMBINING;
  1;
  ";
  close OUT;
}


sub run_test {
  my $count = shift;

  mkdir ("logs");
  system ("./bayes-analyse-from-raw-counts $SPAMLOG $HAMLOG ".
	    "> logs/log.$count 2>&1");
}

sub parse_log {
  my $rands = shift;
  my $count = shift;
  my $r = { };

  open (IN, "< logs/log.$count");
  while (<IN>) {
    /optimization for hamcutoff=(\S+), spamcutoff=(\S+): cost=\$\s*(\S+)/ or next;
    $r->{hamcutoff} = $1;
    $r->{spamcutoff} = $2;
    $r->{cost} = $3;

    $_ = <IN>;
    $_ = <IN>; s/\s+/ /gs; /FP: (\S+) \S+ FN: (\S+) /;
    $r->{fp} = $1;
    $r->{fn} = $2;

    $_ = <IN>; s/\s+/ /gs; /Unsure: (\S+) \S+ .ham: (\S+) \S+ spam: (\S+) /;
    $r->{unsure} = $1;
    $r->{unham} = $2;
    $r->{unspam} = $3;

    last;
  }
  close IN;

  my $line = "$count";
  foreach my $key (sort keys %{$rands}) {
    $line .= " $key $rands->{$key}";
  }
  foreach my $key (sort keys %{$r}) {
    $line .= " $key $r->{$key}";
  }
  print $line, "\n";
  sleep 1;
}