MySQL Query - 使用 Aggregate 和 Group By 生成单独的结果

Question

我有一个表，其中一个或多个具有相同“id”值的条目可以插入到我们的日志/事实表中（包含超过 100+ 百万条记录）
以设定的频率将一条新记录插入到此表中，并为“created”和“view_percent”列（观看视频的百分比）设置一个新值。
对于两个不同的查询，我想返回：

预期结果 1：

+------------------+--------------+-------------+------------------+------------+ | archive_asset_id | asset_title | count_asset | avg_view | time_day | +------------------+--------------+-------------+------------------+------------+ | 83386 | Oliver James | 4 | 36.75 | 2010-08-09 | +------------------+--------------+-------------+------------------+------------+

使用此查询，我无法在执行聚合函数时过滤掉必要的记录...取而代之的是所有行的平均值，结果值为 31.307

SELECT archive_asset_id, asset_title, COUNT(DISTINCT id * 1000000 + archive_asset_id) AS count_asset, AVG(view_percent) AS avg_view, FROM_UNIXTIME(created, '%Y-%m-%d') AS time_day
FROM log_embed_video 
WHERE archive_asset_id = 83386 
AND created >= 1281312000
AND created < 1281484800
GROUP BY time_day
ORDER BY time_day;

预期结果 2：

+------------+------------------+---------------+------------------+-------------------------------+ | time_day | archive_asset_id | asset_title | MAX(view_percent) | occurrences MAX(view_percent) | +------------+------------------+---------------+------------------+-------------------------------+ | 2010-08-09 | 83386 | Oliver James | 13 | 1 | | 2010-08-09 | 83386 | Oliver James | 17 | 2 | | 2010-08-09 | 83386 | Oliver James | 100 | 1 | +------------+------------------+---------------+-------------------+------------------------------+

这是我用于结果 2 的查询，但不是我想要的... log_embed_video.id 的组产生 4 个结果... 这是给定查询的预期结果，但不是预期的输出。

SELECT id, FROM_UNIXTIME(created, '%Y-%m-%d') AS time_day, archive_asset_id, asset_title, COUNT(DISTINCT id * 1000000 + archive_asset_id) AS 'count_asset', MAX(view_percent) as 'max_view_percent'
FROM log_embed_video 
WHERE archive_asset_id = 83386 
AND created >= 1281312000
AND created < 1281484800
GROUP BY time_day, id

条件数据：
带有 KEEP 标记的行是我在返回结果 1 和结果 2 时要使用的数据。

SELECT id, archive_asset_id, asset_title, view_percent, FROM_UNIXTIME(created, '%Y-%m-%d') AS time_day
FROM log_embed_video 
WHERE archive_asset_id = 83386 
AND created >= 1281312000
AND created < 1281484800
ORDER BY id, view_percent;

+----------+------------------+--------------+--------------+------------+ | id | archive_asset_id | asset_title | view_percent | time_day | +----------+------------------+--------------+--------------+------------+ | 43326898 | 83386 | Oliver James | 0 | 2010-08-09 | - DISCARD RECORD / DUPLICATE | 43326898 | 83386 | Oliver James | 13 | 2010-08-09 | + KEEP | 43432090 | 83386 | Oliver James | 0 | 2010-08-09 | - DISCARD RECORD / DUPLICATE | 43432090 | 83386 | Oliver James | 17 | 2010-08-09 | + KEEP | 43432092 | 83386 | Oliver James | 0 | 2010-08-09 | - DISCARD RECORD / DUPLICATE | 43432092 | 83386 | Oliver James | 17 | 2010-08-09 | + KEEP | 43470093 | 83386 | Oliver James | 0 | 2010-08-09 | - DISCARD RECORD / DUPLICATE | 43470093 | 83386 | Oliver James | 17 | 2010-08-09 | - DISCARD RECORD / DUPLICATE | 43470093 | 83386 | Oliver James | 35 | 2010-08-09 | - DISCARD RECORD / DUPLICATE | 43470093 | 83386 | Oliver James | 52 | 2010-08-09 | - DISCARD RECORD / DUPLICATE | 43470093 | 83386 | Oliver James | 69 | 2010-08-09 | - DISCARD RECORD / DUPLICATE | 43470093 | 83386 | Oliver James | 87 | 2010-08-09 | - DISCARD RECORD / DUPLICATE | 43470093 | 83386 | Oliver James | 100 | 2010-08-09 | + KEEP +----------+------------------+--------------+--------------+------------+

表格和原始数据：

CREATE TABLE `log_embed_video` (
  `id` int(11) NOT NULL,
  `archive_asset_id` int(11) NOT NULL,
  `asset_title` varchar(255) NOT NULL,
    `view_percent` float NOT NULL,
  `created` int(11) NOT NULL
) ENGINE=InnoDB DEFAULT CHARSET=utf8;


INSERT INTO `log_embed_video` VALUES 
(43326898, 83386, 'Oliver James', 0, 1281327306),
(43326898, 83386, 'Oliver James', 13, 1281327327),
(43432090, 83386, 'Oliver James', 0, 1281371423),
(43432090, 83386, 'Oliver James', 17, 1281371445),
(43432092, 83386, 'Oliver James', 0, 1281371424),
(43432092, 83386, 'Oliver James', 17, 1281371446),
(43470093, 83386, 'Oliver James', 0, 1281380789),
(43470093, 83386, 'Oliver James', 17, 1281380810),
(43470093, 83386, 'Oliver James', 35, 1281380830),
(43470093, 83386, 'Oliver James', 52, 1281380850),
(43470093, 83386, 'Oliver James', 69, 1281380871),
(43470093, 83386, 'Oliver James', 87, 1281380891),
(43470093, 83386, 'Oliver James', 100, 1281380906);

Answer 1

唯一 id 的所有 max-percentage per-id 行：

SELECT a.* 
FROM log_embed_video a 
LEFT JOIN log_embed_video b
ON b.id = a.id
AND b.view_percent > a.view_percent
WHERE b.id IS NULL
-- possibly limit on date for  more performance.

在性能方面这更好：

SELECT * FROM (
    SELECT id, archive_asset_id, asset_title, view_percent, created,
        @rn := IF(id != @old_id,1,@rn + 1) as rownumber,
        @old_id := id 
    FROM log_embed_video 
    JOIN (SELECT @rn:=0,@old_id:=0) void
    ORDER BY id, view_percent DESC
) a WHERE rownumber=1;

Answer 2

检查这是否会让你更清楚

SELECT archive_asset_id, AVG(actual_percent) 
FROM (SELECT id, archive_asset_id, asset_title, 
             MAX(view_percent) as actual_percent 
      FROM log_embed_video GROUP by id) T 
GROUP BY archive_asset_id;

返回：

+------------------+---------------------+
| archive_asset_id | AVG(actual_percent) |
+------------------+---------------------+
|            83386 |               36.75 | 
+------------------+---------------------+

一些笔记

• 这在 1 亿条记录上表现不佳
• 您还可能希望对数据进行规范化以提高性能（在这种情况下会这样做；基本上将实际的最终行移到它们自己的表中对我来说更有意义）
• COUNT(DISTINCT id * 1000000 + archive_asset_id) 这个表达式引起了我的注意，因为它很奇怪；你确定你的意思不是简单的COUNT(*) 或COUNT(id)？

编辑：

第二个

SELECT archive_asset_id, actual_percent, count(*) 
FROM (SELECT id, archive_asset_id, asset_title,               
             MAX(view_percent) as actual_percent        
      FROM log_embed_video GROUP by id) T  
GROUP BY archive_asset_id, actual_percent;

+------------------+----------------+----------+
| archive_asset_id | actual_percent | count(*) |
+------------------+----------------+----------+
|            83386 |             13 |        1 | 
|            83386 |             17 |        2 | 
|            83386 |            100 |        1 | 
+------------------+----------------+----------+