我希望在一系列日期中返回一周中一天中的小时的滚动/移动平均值，过去几周的范围。
范围内每个日期的回顾都有一个偏移和一个跨周。对于范围中的每个日期，我想从前几周中提取值，使用一周的开始周数，以及过去一周的数周。
例如，10月1日 - 10月7日，回顾4周，从1周开始。
10月1日至10月7日的每一天，回顾4周的时间，从一周开始，并记录每个小时的计数总和，每天的同一天（1-7） ，在回顾范围内。
然后我们可以用它来做一些数学计算，在这种情况下，得到平均入场密度，或者该范围内每个日期的每周每小时的平均入场次数。
从该范围内的10月1日开始。
这应该产生24行，包括：
日期栏填写于10月1日（2017-10-01），
一天的专栏，（周日1），
一个计数列，每个小时桶的条目总数，对于每个dayofweek（1）的实例，在回溯范围内（4周）
在10月1日返回的一系列星期日（星期日）是（尤达税）。
因此，需要从9月3日，10日，17日和24日开始每小时的参赛作品数，并计算当天每个小时的平均数，超过这4个星期日。 10月1日上午10点的响应值将是3日，10日，17日和24日上午10点的行数之和。
那是10月1日。
对于该范围内的10月2日，它将拉出24行，日期列为2017-10-02，dayofweek列（星期一为2），每个小时的平均距离为回溯距离，其中包括： 9月4日， 9月11日， 9月18日
选择HOUR（日期栏）的计数，从每一天开始，总和将除以每个小时段具有值的回顾范围内的天数。
架构设置：
CREATE TABLE `myTable` (
`datetimecolumn` datetime<br/>
)
INSERT INTO MyTable
(`datetimecolumn`)
VALUES
('2015-08-01 00:01:00'),
('2015-08-01 00:21:00'),
('2015-08-01 01:25:00'),
('2015-08-01 01:39:00'),
('2015-08-01 02:11:00'),
('2015-08-01 03:01:00'),
## more here, several randomly timed entries in most hours of each day
('2017-10-16 01:01:00'),
('2017-10-16 02:11:00'),
('2017-10-16 02:21:00'),
('2017-10-16 03:01:00'),
('2017-10-16 05:43:00'),
('2017-10-16 06:21:00')
;
2017-09-30至2017-10-02所需结果：
| Date | DOW | hour_of_day | Avg_Num_Entries |
|------------|-----|-------------| ----------------|
| 2017-09-30 | 7 | 00 | 12 |
| 2017-09-30 | 7 | 01 | 11 |
| 2017-09-30 | 7 | 02 | 14 |
| 2017-09-30 | 7 | 03 | 12 |
| 2017-09-30 | 7 | 04 | 11 |
| 2017-09-30 | 7 | 05 | 14 |
( ..24 total rows for DAYOFWEEK 7 )
| 2017-10-01 | 1 | 00 | 12 |
| 2017-10-01 | 1 | 01 | 11 |
| 2017-10-01 | 1 | 02 | 14 |
| 2017-10-01 | 1 | 03 | 12 |
| 2017-10-01 | 1 | 04 | 11 |
| 2017-10-01 | 1 | 05 | 14 |
( ..24 total rows for DAYOFWEEK 1 )
| 2017-10-02 | 2 | 00 | 12 |
| 2017-10-02 | 2 | 01 | 11 |
| 2017-10-02 | 2 | 02 | 14 |
| 2017-10-02 | 2 | 03 | 12 |
| 2017-10-02 | 2 | 04 | 11 |
| 2017-10-02 | 2 | 05 | 14 |
( ..24 total rows for DAYOFWEEK 2 )
我目前正在处理这个问题：
SELECT
DATE(c.datetimecolum) datetimecolum,
t.sq_datetimecolum,
t.sq_hour,
t.sq_count
FROM
calls c
INNER JOIN(
SELECT
DATE(datetimecolum) sq_datetimecolum,
DAYOFWEEK(datetimecolum) sq_dow,
HOUR(datetimecolum) sq_hour,
COUNT(*) sq_count
FROM
calls
GROUP BY sq_datetimecolum,sq_dow,sq_hour
) AS t
ON
DATE(t.sq_datetimecolum) <= DATE_SUB(
DATE(c.datetimecolum),
INTERVAL 1 WEEK
) AND DATE(t.sq_datetimecolum) >= DATE_SUB(
DATE(c.datetimecolum),
INTERVAL 4 WEEK
) AND DAYOFWEEK(t.sq_datetimecolum) = DAYOFWEEK(DATE(c.datetimecolum))
WHERE DATE(c.datetimecolum) >= '2017-08-01' AND DATE(c.datetimecolum) <= '2017-08-02'
GROUP BY DATE(c.datetimecolum),DATE(t.sq_datetimecolum),DAYOFWEEK(t.sq_datetimecolum),t.sq_hour
ORDER BY DATE(c.datetimecolum),t.sq_datetimecolum,t.sq_hour
为2017-08-01至2017-08-02的日期范围生成此输出：
| datetimecolum | sq_datetimecolum | sq_hour | sq_count |
|----------------------|------------------|---------|----------|
| 2017-08-01 | 2017-07-04 | 0 | 18 |
| 2017-08-01 | 2017-07-04 | 1 | 11 |
| 2017-08-01 | 2017-07-04 | 2 | 7 |
| 2017-08-01 | 2017-07-04 | 3 | 2 |
| 2017-08-01 | 2017-07-04 | 4 | 3 |
| 2017-08-01 | 2017-07-04 | 5 | 2 |
| 2017-08-01 | 2017-07-04 | 6 | 4 |
| 2017-08-01 | 2017-07-04 | 7 | 1 |
| 2017-08-01 | 2017-07-04 | 8 | 4 |
| 2017-08-01 | 2017-07-04 | 9 | 6 |
| 2017-08-01 | 2017-07-04 | 10 | 6 |
| 2017-08-01 | 2017-07-04 | 11 | 3 |
| 2017-08-01 | 2017-07-04 | 12 | 11 |
| 2017-08-01 | 2017-07-04 | 13 | 21 |
| 2017-08-01 | 2017-07-04 | 14 | 44 |
| 2017-08-01 | 2017-07-04 | 15 | 52 |
| 2017-08-01 | 2017-07-04 | 16 | 32 |
| 2017-08-01 | 2017-07-04 | 17 | 26 |
| 2017-08-01 | 2017-07-04 | 18 | 22 |
| 2017-08-01 | 2017-07-04 | 19 | 25 |
| 2017-08-01 | 2017-07-04 | 20 | 14 |
| 2017-08-01 | 2017-07-04 | 21 | 7 |
| 2017-08-01 | 2017-07-04 | 22 | 10 |
| 2017-08-01 | 2017-07-04 | 23 | 6 |
| 2017-08-01 | 2017-07-11 | 0 | 21 |
| 2017-08-01 | 2017-07-11 | 1 | 21 |
| 2017-08-01 | 2017-07-11 | 2 | 12 |
| 2017-08-01 | 2017-07-11 | 3 | 12 |
| 2017-08-01 | 2017-07-11 | 4 | 3 |
| 2017-08-01 | 2017-07-11 | 5 | 4 |
| 2017-08-01 | 2017-07-11 | 6 | 2 |
| 2017-08-01 | 2017-07-11 | 7 | 3 |
| 2017-08-01 | 2017-07-11 | 8 | 7 |
| 2017-08-01 | 2017-07-11 | 10 | 5 |
| 2017-08-01 | 2017-07-11 | 11 | 14 |
| 2017-08-01 | 2017-07-11 | 12 | 50 |
| 2017-08-01 | 2017-07-11 | 13 | 99 |
| 2017-08-01 | 2017-07-11 | 14 | 117 |
| 2017-08-01 | 2017-07-11 | 15 | 132 |
| 2017-08-01 | 2017-07-11 | 16 | 101 |
| 2017-08-01 | 2017-07-11 | 17 | 89 |
| 2017-08-01 | 2017-07-11 | 18 | 81 |
| 2017-08-01 | 2017-07-11 | 19 | 110 |
| 2017-08-01 | 2017-07-11 | 20 | 120 |
| 2017-08-01 | 2017-07-11 | 21 | 119 |
| 2017-08-01 | 2017-07-11 | 22 | 85 |
| 2017-08-01 | 2017-07-11 | 23 | 35 |
| 2017-08-01 | 2017-07-18 | 0 | 44 |
| 2017-08-01 | 2017-07-18 | 1 | 33 |
| 2017-08-01 | 2017-07-18 | 2 | 11 |
| 2017-08-01 | 2017-07-18 | 3 | 13 |
| 2017-08-01 | 2017-07-18 | 4 | 7 |
| 2017-08-01 | 2017-07-18 | 5 | 2 |
| 2017-08-01 | 2017-07-18 | 6 | 7 |
| 2017-08-01 | 2017-07-18 | 7 | 5 |
| 2017-08-01 | 2017-07-18 | 8 | 7 |
| 2017-08-01 | 2017-07-18 | 9 | 10 |
| 2017-08-01 | 2017-07-18 | 10 | 5 |
| 2017-08-01 | 2017-07-18 | 11 | 16 |
| 2017-08-01 | 2017-07-18 | 12 | 39 |
| 2017-08-01 | 2017-07-18 | 13 | 67 |
| 2017-08-01 | 2017-07-18 | 14 | 128 |
| 2017-08-01 | 2017-07-18 | 15 | 133 |
| 2017-08-01 | 2017-07-18 | 16 | 129 |
| 2017-08-01 | 2017-07-18 | 17 | 120 |
| 2017-08-01 | 2017-07-18 | 18 | 102 |
| 2017-08-01 | 2017-07-18 | 19 | 93 |
| 2017-08-01 | 2017-07-18 | 20 | 112 |
| 2017-08-01 | 2017-07-18 | 21 | 81 |
| 2017-08-01 | 2017-07-18 | 22 | 74 |
| 2017-08-01 | 2017-07-18 | 23 | 43 |
| 2017-08-01 | 2017-07-25 | 0 | 32 |
| 2017-08-01 | 2017-07-25 | 1 | 19 |
| 2017-08-01 | 2017-07-25 | 2 | 9 |
| 2017-08-01 | 2017-07-25 | 3 | 3 |
| 2017-08-01 | 2017-07-25 | 4 | 6 |
| 2017-08-01 | 2017-07-25 | 5 | 8 |
| 2017-08-01 | 2017-07-25 | 6 | 3 |
| 2017-08-01 | 2017-07-25 | 7 | 2 |
| 2017-08-01 | 2017-07-25 | 8 | 3 |
| 2017-08-01 | 2017-07-25 | 9 | 2 |
| 2017-08-01 | 2017-07-25 | 10 | 5 |
| 2017-08-01 | 2017-07-25 | 11 | 15 |
| 2017-08-01 | 2017-07-25 | 12 | 39 |
| 2017-08-01 | 2017-07-25 | 13 | 71 |
| 2017-08-01 | 2017-07-25 | 14 | 84 |
| 2017-08-01 | 2017-07-25 | 15 | 100 |
| 2017-08-01 | 2017-07-25 | 16 | 92 |
| 2017-08-01 | 2017-07-25 | 17 | 58 |
| 2017-08-01 | 2017-07-25 | 18 | 59 |
| 2017-08-01 | 2017-07-25 | 19 | 70 |
| 2017-08-01 | 2017-07-25 | 20 | 80 |
| 2017-08-01 | 2017-07-25 | 21 | 83 |
| 2017-08-01 | 2017-07-25 | 22 | 60 |
| 2017-08-01 | 2017-07-25 | 23 | 40 |
| 2017-08-02 | 2017-07-05 | 0 | 7 |
| 2017-08-02 | 2017-07-05 | 1 | 6 |
| 2017-08-02 | 2017-07-05 | 2 | 4 |
| 2017-08-02 | 2017-07-05 | 3 | 4 |
| 2017-08-02 | 2017-07-05 | 4 | 3 |
| 2017-08-02 | 2017-07-05 | 5 | 1 |
| 2017-08-02 | 2017-07-05 | 6 | 2 |
| 2017-08-02 | 2017-07-05 | 7 | 1 |
| 2017-08-02 | 2017-07-05 | 8 | 3 |
| 2017-08-02 | 2017-07-05 | 9 | 3 |
| 2017-08-02 | 2017-07-05 | 10 | 5 |
| 2017-08-02 | 2017-07-05 | 11 | 13 |
| 2017-08-02 | 2017-07-05 | 12 | 50 |
| 2017-08-02 | 2017-07-05 | 13 | 97 |
| 2017-08-02 | 2017-07-05 | 14 | 125 |
| 2017-08-02 | 2017-07-05 | 15 | 139 |
| 2017-08-02 | 2017-07-05 | 16 | 106 |
| 2017-08-02 | 2017-07-05 | 17 | 106 |
| 2017-08-02 | 2017-07-05 | 18 | 93 |
| 2017-08-02 | 2017-07-05 | 19 | 80 |
| 2017-08-02 | 2017-07-05 | 20 | 86 |
| 2017-08-02 | 2017-07-05 | 21 | 68 |
| 2017-08-02 | 2017-07-05 | 22 | 53 |
| 2017-08-02 | 2017-07-05 | 23 | 26 |
| 2017-08-02 | 2017-07-12 | 0 | 24 |
| 2017-08-02 | 2017-07-12 | 1 | 21 |
| 2017-08-02 | 2017-07-12 | 2 | 16 |
| 2017-08-02 | 2017-07-12 | 3 | 3 |
| 2017-08-02 | 2017-07-12 | 4 | 4 |
| 2017-08-02 | 2017-07-12 | 5 | 5 |
| 2017-08-02 | 2017-07-12 | 6 | 1 |
| 2017-08-02 | 2017-07-12 | 7 | 2 |
| 2017-08-02 | 2017-07-12 | 8 | 5 |
| 2017-08-02 | 2017-07-12 | 9 | 15 |
| 2017-08-02 | 2017-07-12 | 10 | 7 |
| 2017-08-02 | 2017-07-12 | 11 | 12 |
| 2017-08-02 | 2017-07-12 | 12 | 44 |
| 2017-08-02 | 2017-07-12 | 13 | 60 |
| 2017-08-02 | 2017-07-12 | 14 | 73 |
| 2017-08-02 | 2017-07-12 | 15 | 119 |
| 2017-08-02 | 2017-07-12 | 16 | 100 |
| 2017-08-02 | 2017-07-12 | 17 | 87 |
| 2017-08-02 | 2017-07-12 | 18 | 58 |
| 2017-08-02 | 2017-07-12 | 19 | 85 |
| 2017-08-02 | 2017-07-12 | 20 | 96 |
| 2017-08-02 | 2017-07-12 | 21 | 79 |
| 2017-08-02 | 2017-07-12 | 22 | 61 |
| 2017-08-02 | 2017-07-12 | 23 | 25 |
| 2017-08-02 | 2017-07-19 | 0 | 27 |
| 2017-08-02 | 2017-07-19 | 1 | 24 |
| 2017-08-02 | 2017-07-19 | 2 | 20 |
| 2017-08-02 | 2017-07-19 | 3 | 12 |
| 2017-08-02 | 2017-07-19 | 4 | 4 |
| 2017-08-02 | 2017-07-19 | 5 | 4 |
| 2017-08-02 | 2017-07-19 | 6 | 2 |
| 2017-08-02 | 2017-07-19 | 7 | 5 |
| 2017-08-02 | 2017-07-19 | 8 | 4 |
| 2017-08-02 | 2017-07-19 | 9 | 4 |
| 2017-08-02 | 2017-07-19 | 10 | 10 |
| 2017-08-02 | 2017-07-19 | 11 | 11 |
| 2017-08-02 | 2017-07-19 | 12 | 33 |
| 2017-08-02 | 2017-07-19 | 13 | 75 |
| 2017-08-02 | 2017-07-19 | 14 | 107 |
| 2017-08-02 | 2017-07-19 | 15 | 98 |
| 2017-08-02 | 2017-07-19 | 16 | 68 |
| 2017-08-02 | 2017-07-19 | 17 | 71 |
| 2017-08-02 | 2017-07-19 | 18 | 96 |
| 2017-08-02 | 2017-07-19 | 19 | 102 |
| 2017-08-02 | 2017-07-19 | 20 | 86 |
| 2017-08-02 | 2017-07-19 | 21 | 67 |
| 2017-08-02 | 2017-07-19 | 22 | 54 |
| 2017-08-02 | 2017-07-19 | 23 | 29 |
| 2017-08-02 | 2017-07-26 | 0 | 35 |
| 2017-08-02 | 2017-07-26 | 1 | 16 |
| 2017-08-02 | 2017-07-26 | 2 | 11 |
| 2017-08-02 | 2017-07-26 | 3 | 8 |
| 2017-08-02 | 2017-07-26 | 4 | 6 |
| 2017-08-02 | 2017-07-26 | 5 | 5 |
| 2017-08-02 | 2017-07-26 | 6 | 3 |
| 2017-08-02 | 2017-07-26 | 7 | 3 |
| 2017-08-02 | 2017-07-26 | 8 | 7 |
| 2017-08-02 | 2017-07-26 | 9 | 2 |
| 2017-08-02 | 2017-07-26 | 10 | 5 |
| 2017-08-02 | 2017-07-26 | 11 | 11 |
| 2017-08-02 | 2017-07-26 | 12 | 41 |
| 2017-08-02 | 2017-07-26 | 13 | 52 |
| 2017-08-02 | 2017-07-26 | 14 | 76 |
| 2017-08-02 | 2017-07-26 | 15 | 83 |
| 2017-08-02 | 2017-07-26 | 16 | 97 |
| 2017-08-02 | 2017-07-26 | 17 | 71 |
| 2017-08-02 | 2017-07-26 | 18 | 56 |
| 2017-08-02 | 2017-07-26 | 19 | 69 |
| 2017-08-02 | 2017-07-26 | 20 | 45 |
| 2017-08-02 | 2017-07-26 | 21 | 67 |
| 2017-08-02 | 2017-07-26 | 22 | 68 |
| 2017-08-02 | 2017-07-26 | 23 | 39 |
我很亲密。这将产生每周一天的每小时数量，从4周开始。
我只需要在4天内将这些小时总数相加，然后除以4。
答案 0 :(得分：1)
此方法使用2个子查询：
子查询1 通过在指定日期范围内形成日期，星期几和小时的唯一行来控制最终输出。该日期范围使用该范围的起始日期的参数，即在下一个子查询中使用相同的参数。 （ nb：这里假设所有必需的天数和小时数将出现在数据中。如果没有，则需要另一种方法来提供所需的所有天数和小时数。
子查询2 从上面使用的起点向后过滤，可以通过参数控制若干周。然后按日周和小时汇总此日期范围。
然后在每周和每小时加入2个子查询，以便平均值与ate，dayofweek和hour对齐。请注意，每个子查询中使用的日期范围是互斥的，因此计数是准确的。
MySQL 5.6架构设置：
CREATE TABLE `myTable` (
`id` mediumint(8) unsigned NOT NULL auto_increment,
`start_time` datetime,
PRIMARY KEY (`id`)
) AUTO_INCREMENT=1;
INSERT INTO MyTable
(`start_time`)
VALUES
('2017-08-05 00:01:00'),
('2017-08-05 00:15:00'),
('2017-08-05 00:29:00'),
('2017-08-05 01:01:00'),
('2017-08-05 01:15:00'),
('2017-08-05 01:29:00'),
('2017-08-05 02:01:00'),
('2017-08-05 02:15:00'),
('2017-08-05 02:29:00'),
('2017-08-05 03:01:00'),
('2017-08-05 03:15:00'),
('2017-08-05 03:29:00'),
('2017-08-05 04:01:00'),
('2017-08-05 04:15:00'),
('2017-08-05 04:29:00'),
('2017-08-05 05:01:00'),
('2017-08-05 05:15:00'),
('2017-08-05 05:29:00'),
('2017-08-05 06:01:00'),
('2017-08-05 06:15:00'),
('2017-08-05 06:29:00'),
('2017-08-05 07:01:00'),
('2017-08-05 07:15:00'),
('2017-08-05 07:29:00'),
('2017-08-05 08:01:00'),
('2017-08-05 08:15:00'),
('2017-08-05 08:29:00'),
('2017-08-05 09:01:00'),
('2017-08-05 09:15:00'),
('2017-08-05 09:29:00'),
('2017-08-05 10:01:00'),
('2017-08-05 10:15:00'),
('2017-08-05 10:29:00'),
('2017-08-05 11:01:00'),
('2017-08-05 11:15:00'),
('2017-08-05 11:29:00'),
('2017-08-05 12:01:00'),
('2017-08-05 12:15:00'),
('2017-08-05 12:29:00'),
('2017-08-05 13:01:00'),
('2017-08-05 13:15:00'),
('2017-08-05 13:29:00'),
('2017-08-05 14:01:00'),
('2017-08-05 14:15:00'),
('2017-08-05 14:29:00'),
('2017-08-05 15:01:00'),
('2017-08-05 15:15:00'),
('2017-08-05 15:29:00'),
('2017-08-05 16:01:00'),
('2017-08-05 16:15:00'),
('2017-08-05 16:29:00'),
('2017-08-05 17:01:00'),
('2017-08-05 17:15:00'),
('2017-08-05 17:29:00'),
('2017-08-05 18:01:00'),
('2017-08-05 18:15:00'),
('2017-08-05 18:29:00'),
('2017-08-05 19:01:00'),
('2017-08-05 19:15:00'),
('2017-08-05 19:29:00'),
('2017-08-05 20:01:00'),
('2017-08-05 20:15:00'),
('2017-08-05 20:29:00'),
('2017-08-05 21:01:00'),
('2017-08-05 21:15:00'),
('2017-08-05 21:29:00'),
('2017-08-05 22:01:00'),
('2017-08-05 22:15:00'),
('2017-08-05 22:29:00'),
('2017-08-05 23:01:00'),
('2017-08-05 23:15:00'),
('2017-08-05 23:29:00'),
('2017-08-06 00:01:00'),
('2017-08-06 00:15:00'),
('2017-08-06 00:29:00'),
('2017-08-06 01:01:00'),
('2017-08-06 01:15:00'),
('2017-08-06 01:29:00'),
('2017-08-06 02:01:00'),
('2017-08-06 02:15:00'),
('2017-08-06 02:29:00'),
('2017-08-06 03:01:00'),
('2017-08-06 03:15:00'),
('2017-08-06 03:29:00'),
('2017-08-06 04:01:00'),
('2017-08-06 04:15:00'),
('2017-08-06 04:29:00'),
('2017-08-06 05:01:00'),
('2017-08-06 05:15:00'),
('2017-08-06 05:29:00'),
('2017-08-06 06:01:00'),
('2017-08-06 06:15:00'),
('2017-08-06 06:29:00'),
('2017-08-06 07:01:00'),
('2017-08-06 07:15:00'),
('2017-08-06 07:29:00'),
('2017-08-06 08:01:00'),
('2017-08-06 08:15:00'),
('2017-08-06 08:29:00'),
('2017-08-06 09:01:00'),
('2017-08-06 09:15:00'),
('2017-08-06 09:29:00'),
('2017-08-06 10:01:00'),
('2017-08-06 10:15:00'),
('2017-08-06 10:29:00'),
('2017-08-06 11:01:00'),
('2017-08-06 11:15:00'),
('2017-08-06 11:29:00'),
('2017-08-06 12:01:00'),
('2017-08-06 12:15:00'),
('2017-08-06 12:29:00'),
('2017-08-06 13:01:00'),
('2017-08-06 13:15:00'),
('2017-08-06 13:29:00'),
('2017-08-06 14:01:00'),
('2017-08-06 14:15:00'),
('2017-08-06 14:29:00'),
('2017-08-06 15:01:00'),
('2017-08-06 15:15:00'),
('2017-08-06 15:29:00'),
('2017-08-06 16:01:00'),
('2017-08-06 16:15:00'),
('2017-08-06 16:29:00'),
('2017-08-06 17:01:00'),
('2017-08-06 17:15:00'),
('2017-08-06 17:29:00'),
('2017-08-06 18:01:00'),
('2017-08-06 18:15:00'),
('2017-08-06 18:29:00'),
('2017-08-06 19:01:00'),
('2017-08-06 19:15:00'),
('2017-08-06 19:29:00'),
('2017-08-06 20:01:00'),
('2017-08-06 20:15:00'),
('2017-08-06 20:29:00'),
('2017-08-06 21:01:00'),
('2017-08-06 21:15:00'),
('2017-08-06 21:29:00'),
('2017-08-06 22:01:00'),
('2017-08-06 22:15:00'),
('2017-08-06 22:29:00'),
('2017-08-06 23:01:00'),
('2017-08-06 23:15:00'),
('2017-08-06 23:29:00'),
('2017-08-03 00:01:00'),
('2017-08-03 00:15:00'),
('2017-08-03 00:29:00'),
('2017-08-03 01:01:00'),
('2017-08-03 01:15:00'),
('2017-08-03 01:29:00'),
('2017-08-03 02:01:00'),
('2017-08-03 02:15:00'),
('2017-08-03 02:29:00'),
('2017-08-03 03:01:00'),
('2017-08-03 03:15:00'),
('2017-08-03 03:29:00'),
('2017-08-03 04:01:00'),
('2017-08-03 04:15:00'),
('2017-08-03 04:29:00'),
('2017-08-03 05:01:00'),
('2017-08-03 05:15:00'),
('2017-08-03 05:29:00'),
('2017-08-03 06:01:00'),
('2017-08-03 06:15:00'),
('2017-08-03 06:29:00'),
('2017-08-03 07:01:00'),
('2017-08-03 07:15:00'),
('2017-08-03 07:29:00'),
('2017-08-03 08:01:00'),
('2017-08-03 08:15:00'),
('2017-08-03 08:29:00'),
('2017-08-12 00:01:00'),
('2017-08-12 00:15:00'),
('2017-08-12 00:29:00'),
('2017-08-12 01:01:00'),
('2017-08-12 01:15:00'),
('2017-08-12 01:29:00'),
('2017-08-12 02:01:00'),
('2017-08-12 02:15:00'),
('2017-08-12 02:29:00'),
('2017-08-12 03:01:00'),
('2017-08-12 03:15:00'),
('2017-08-12 03:29:00'),
('2017-08-12 04:01:00'),
('2017-08-12 04:15:00'),
('2017-08-12 04:29:00'),
('2017-08-12 05:01:00'),
('2017-08-12 05:15:00'),
('2017-08-12 05:29:00'),
('2017-08-12 06:01:00'),
('2017-08-12 06:15:00'),
('2017-08-12 06:29:00'),
('2017-08-12 07:01:00'),
('2017-08-12 07:15:00'),
('2017-08-12 07:29:00'),
('2017-08-12 08:01:00'),
('2017-08-12 08:15:00'),
('2017-08-12 08:29:00'),
('2017-08-12 09:01:00'),
('2017-08-12 09:15:00'),
('2017-08-12 09:29:00'),
('2017-08-12 10:01:00'),
('2017-08-12 10:15:00'),
('2017-08-12 10:29:00'),
('2017-08-12 11:01:00'),
('2017-08-12 11:15:00'),
('2017-08-12 11:29:00'),
('2017-08-12 12:01:00'),
('2017-08-12 12:15:00'),
('2017-08-12 12:29:00'),
('2017-08-12 13:01:00'),
('2017-08-12 13:15:00'),
('2017-08-12 13:29:00'),
('2017-08-12 14:01:00'),
('2017-08-12 14:15:00'),
('2017-08-12 14:29:00'),
('2017-08-12 15:01:00'),
('2017-08-12 15:15:00'),
('2017-08-12 15:29:00'),
('2017-08-12 16:01:00'),
('2017-08-12 16:15:00'),
('2017-08-12 16:29:00'),
('2017-08-12 17:01:00'),
('2017-08-12 17:15:00'),
('2017-08-12 17:29:00'),
('2017-08-12 18:01:00'),
('2017-08-12 18:15:00'),
('2017-08-12 18:29:00'),
('2017-08-12 19:01:00'),
('2017-08-12 19:15:00'),
('2017-08-12 19:29:00'),
('2017-08-12 20:01:00'),
('2017-08-12 20:15:00'),
('2017-08-12 20:29:00'),
('2017-08-12 21:01:00'),
('2017-08-12 21:15:00'),
('2017-08-12 21:29:00'),
('2017-08-12 22:01:00'),
('2017-08-12 22:15:00'),
('2017-08-12 22:29:00'),
('2017-08-12 23:01:00'),
('2017-08-12 23:15:00'),
('2017-08-12 23:29:00'),
('2017-08-13 00:01:00'),
('2017-08-13 00:15:00'),
('2017-08-13 00:29:00'),
('2017-08-13 01:01:00'),
('2017-08-13 01:15:00'),
('2017-08-13 01:29:00'),
('2017-08-13 02:01:00'),
('2017-08-13 02:15:00'),
('2017-08-13 02:29:00'),
('2017-08-13 03:01:00'),
('2017-08-13 03:15:00'),
('2017-08-13 03:29:00'),
('2017-08-13 04:01:00'),
('2017-08-13 04:15:00'),
('2017-08-13 04:29:00'),
('2017-08-13 05:01:00'),
('2017-08-13 05:15:00'),
('2017-08-13 05:29:00'),
('2017-08-13 06:01:00'),
('2017-08-13 06:15:00'),
('2017-08-13 06:29:00'),
('2017-08-13 07:01:00'),
('2017-08-13 07:15:00'),
('2017-08-13 07:29:00'),
('2017-08-13 08:01:00'),
('2017-08-13 08:15:00'),
('2017-08-13 08:29:00'),
('2017-08-13 09:01:00'),
('2017-08-13 09:15:00'),
('2017-08-13 09:29:00'),
('2017-08-13 10:01:00'),
('2017-08-13 10:15:00'),
('2017-08-13 10:29:00'),
('2017-08-13 11:01:00'),
('2017-08-13 11:15:00'),
('2017-08-13 11:29:00'),
('2017-08-13 12:01:00'),
('2017-08-13 12:15:00'),
('2017-08-13 12:29:00'),
('2017-08-13 13:01:00'),
('2017-08-13 13:15:00'),
('2017-08-13 13:29:00'),
('2017-08-13 14:01:00'),
('2017-08-13 14:15:00'),
('2017-08-13 14:29:00'),
('2017-08-13 15:01:00'),
('2017-08-13 15:15:00'),
('2017-08-13 15:29:00'),
('2017-08-13 16:01:00'),
('2017-08-13 16:15:00'),
('2017-08-13 16:29:00'),
('2017-08-13 17:01:00')
;
<强>查询：
set @start_at := '2017-08-12'
set @num_weeks:= 4
set @sub_nw_st := date_sub(@start_at, INTERVAL (7 * @num_weeks) DAY)
select
c.dt, c.dow, c.hr
, coalesce(h.c_count,0) + c.c_count c_count
, coalesce(h.dt_count,0) + 1 dt_count
, (coalesce(h.c_count,0) + c.c_count) / (coalesce(h.dt_count,0) + 1) Avg_Num_Entries
from (
select
date(start_time) dt
, dayofweek(start_time) dow
, hour(start_time) hr
, count(*) c_count
from MyTable
where start_time >= @start_at and start_time < date_add(@start_at,INTERVAL 1 WEEK)
group by
dt, dow, hr
) c
LEFT JOIN (
select
dayofweek(start_time) dow
, hour(start_time) hr
, count(distinct date(start_time)) dt_count
, count(*) c_count
from MyTable
where start_time >= @sub_nw_st and start_time < @start_at
group by
dow, hr
) h
ON c.dow = h.dow and c.hr = h.hr
<强> Results ：
| dt | dow | hr | c_count | dt_count | Avg_Num_Entries |
|------------|-----|----|---------|----------|-----------------|
| 2017-08-12 | 7 | 0 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 1 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 2 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 3 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 4 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 5 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 6 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 7 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 8 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 9 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 10 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 11 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 12 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 13 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 14 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 15 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 16 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 17 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 18 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 19 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 20 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 21 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 22 | 6 | 2 | 3 |
| 2017-08-12 | 7 | 23 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 0 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 1 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 2 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 3 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 4 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 5 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 6 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 7 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 8 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 9 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 10 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 11 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 12 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 13 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 14 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 15 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 16 | 6 | 2 | 3 |
| 2017-08-13 | 1 | 17 | 4 | 2 | 2 |
答案 1 :(得分：0)
更新：这是我目前的解决方案，基于已接受的答案。
SET @start_at := '2017-08-01';
SET @end_at := '2017-08-02';
SET @lookback_offset := 1;
SET @lookback_span := 4;
SELECT
c.*,
count(distinct date(h.dt_col)) days_counted,
count(hour(h.dt_col)) dowhodt,
count(hour(h.dt_col))/count(distinct date(h.dt_col)) dowhodma
FROM
(SELECT
DATE(dt_col) dt,
DAYOFWEEK(dt_col) dow,
HOUR(dt_col) hr
FROM
MyTable
GROUP BY dt , dow , hr) c
LEFT JOIN (
SELECT *
FROM MyTable
) h ON
(DAYOFWEEK(DATE(h.dt_col)) = c.dow)
AND
(HOUR(h.dt_col) = c.hr)
AND
(DATE(h.dt_col) >= DATE_SUB(c.dt, INTERVAL @lookback_span WEEK))
AND
(DATE(h.dt_col) <= DATE_SUB(c.dt, INTERVAL @lookback_offset WEEK))
WHERE DATE( c.dt ) >= @start_at AND DATE( c.dt ) <= @end_at
GROUP BY dt,dow,hr