analyze-retention.php

<?php

$landing = '/';
$goal = '/metric?signup';
$retention = '/metric?signin';

$dir = __DIR__ . '/nginx';
$full = __DIR__ . '/full.log';
unlink($full);
for($i=0;$i<100;$i++) {
    $append = ($i) ? '.' . $i : '';
    $filename = $dir . '/access.log' . $append;
    if (!file_exists($filename)) break;
    echo $filename . PHP_EOL;
    system('cat ' . $filename . ' | grep -v 400 | grep -v 301 | grep -v 302 | grep GET >> ' . $full);
}

$log = array();
$fp = fopen($full, 'r');
while ($line = fgets($fp, 4000)) {
    // 8.8.8.8 - - [01/Mar/2013:15:46:51 +0000] "GET /metric?facebook-page HTTP/1.1" 200 6533 "http://www.truesocialmetrics.com/connect/facebook-page" "Mozilla/5.0 (Macintosh; Intel Mac OS X 10.5; rv:12.0) Gecko/20100101 Firefox/12.0"
    preg_match('~^([0-9\.]+)\s+-\s+-\s+\[([^\]]+)\]\s+"([^"]+)"~', $line, $match);    
    list($method, $uri, $protocol) = explode(' ', $match[3]);

    $log[] = array(
        'ip'  => $match[1],
        'time' => date('Ymd', strtotime($match[2])),
        'uri'  => $uri,
    );
}
$log = array_reverse($log);

$ips = array();
foreach ($log as $item) {
    $ip = $item['ip'];
    if (!array_key_exists($ip, $ips)) {
        $ips[$ip] = array(
            'items' => array(), 
            'step' => 0, 
            'goal_at' => 0,
            'retention_at' => 0,
        );
    }
    $ips[$ip]['items'][] = $item;
}


// step goal calculation
$closure = function($value) use ($goal, $retention) {
    $step = 0;
    $time = 0;
    $successGoal = false;
    foreach ($value['items'] as $item) {
        if ($item['uri']==$goal) {
            $successGoal = true;
            $time = $item['time'];
            break;
        }
        $step++;
    }
    if ($successGoal) {
        $value['step'] = $step;
        $value['goal_at'] = $time;
    }
    return $value;
};

$ips = array_map($closure, $ips);

$ips = array_filter($ips, function($value){
    return $value['step']>0;
});

$closure = function($value) use ($retention) {
    $time = 0;
    foreach ($value['items'] as $item) {
        if ($item['uri']!=$retention) continue;
        $time = $item['time'];
    }
    $value['retention_at'] = $time;
    return $value;
};

$ips = array_map($closure, $ips);
$ips = array_filter($ips, function($value) {
    return $value['retention_at']>0;
});

echo 'step' . "\t" . 'retention' . PHP_EOL;
$steps = array();
$steps = array();
foreach ($ips as $value) {
    $step = $value['step'];
    if (!array_key_exists($step, $steps)) {
        $steps[$step] = array(
            'bounced' => 0,
            'retention' => 0,
            'all' => 0,
        );
    }
    if ($value['retention_at']>0 && $value['retention_at']-$value['goal_at']>1) {
        $steps[$step]['retention']++;
    } else {
        $steps[$step]['bounced']++;        
    }
    $steps[$step]['all']++;
}
ksort($steps);
$groups = array();
foreach ($steps as $step => $value) {
    if (!$value['retention']) continue;
    $rate = $value['retention']/$value['all']*100;
    $groups[$step] = $rate;
}

foreach(array_chunk($groups, 5, true) as $list) {
    $step = round(array_sum(array_keys($list))/count($list));
    $rate = array_sum(array_values($list))/count($list);
    echo $step . "\t" . round($rate) . PHP_EOL;
}