这真的是多个问题，我为问题超载道歉，但我只需要优雅地完成这些。我可以在MySQL中处理简单查询，但是那些复杂的表格通常很困难，而且我还不熟悉动态SQL。寻找简单的解决方案（但不是硬编码）：我不确定在SO的问题中是否要求太多，如果确实太多，请回答一个或两个聚合并给我工具，这样我就可以自己构建这些聚合。
我的数据结构如下：
+-----------------------------------------------------------------+
| timestamp group url metric columns here |
+-----------------------------------------------------------------+
| 2018-05-01 14:30:00 6732 abc.com -0.3673 -0.0914 4.0183 |
| 2018-05-01 14:30:00 6732 xyz.com 4.2187 0.3407 12.3832 |
| 2018-05-01 14:30:00 6732 pqr.org -2.3875 -0.4064 5.8743 |
| 2018-05-01 14:30:00 6732 many.com -4.4194 -1.0665 4.144 |
| 2018-05-01 14:00:00 7174 abc.com -6.4021 -1.419 4.5117 |
| 2018-05-01 14:00:00 7174 xyz.com -1.7971 -1.0396 1.7286 |
| 2018-05-01 14:00:00 7174 many.com 0.5276 0.2621 2.013 |
| 2018-05-01 13:30:00 7174 many.com -0.4941 -0.1098 4.4982 |
| 2018-05-01 13:30:00 7184 diff.com -0.6783 -0.1384 4.9013 |
| 2018-05-01 13:30:00 7184 sites.com -0.1293 -0.0246 5.2608 |
| 2018-05-01 13:30:00 7184 here.com -0.2703 -0.0669 4.0377 |
+-----------------------------------------------------------------+
基本上，对于每个时间戳，我们都有来自不同组的数据，对于每个组，我们都有网址，对于每个网址，我们都会捕获指标。网址和广告组有很多关系。
我必须根据具体情况以多种方式提取和汇总这些数据。通常，我选择我需要的任何指标，并按时间戳，组和网址中的一个或多个进行分组。但是，有时我想在组中看到数据/聚合，但我最终会为它们运行单独的查询。例如，我在特定时间窗口看到的时间聚合，某个指标已经下降或者上升，我想分别深入到每个时间窗口，我将不得不重复这个，因为在一个时间窗口内，某些群体可以上下来挖掘它们以获得url阶段需要一个单独的查询。我需要的是一种在最高级别聚合的方法 - 时间戳和组，但也显示来自以下级别的聚合。一个例子：
这样的事情会有所帮助：
+---------------------+-------------+-------------+------------------+------------------------------+------------------------------+--------------------+--------------------------------+--------------------------------+---------------------+---------------------------------+---------------------------------+
| timestamp | aggregate_1 | aggregate_2 | window_top_group | window_top_group_aggregate_1 | window_top_group_aggregate_2 | window_top_group_2 | window_top_group_2_aggregate_1 | window_top_group_2_aggregate_2 | window_loss_group_1 | window_loss_group_1_aggregate_1 | window_loss_group_1_aggregate_2 |
+---------------------+-------------+-------------+------------------+------------------------------+------------------------------+--------------------+--------------------------------+--------------------------------+---------------------+---------------------------------+---------------------------------+
| 2018-05-01 14:30:00 | -0.3673 | -0.0914 | 6732 | -0.3673 | -0.3673 | 7174 | -0.3673 | -0.3673 | 7184 | -0.3673 | -0.3673 |
| 2018-05-01 14:00:00 | 4.2187 | 0.3407 | 6732 | 4.2187 | 4.2187 | 7174 | 4.2187 | 4.2187 | 7184 | 4.2187 | 4.2187 |
| 2018-05-01 13:30:00 | -2.3875 | -0.4064 | 6732 | -2.3875 | -2.3875 | 7174 | -2.3875 | -2.3875 | 7184 | -2.3875 | -2.3875 |
| 2018-05-01 13:00:00 | -4.4194 | -1.0665 | 6732 | -4.4194 | -4.4194 | 7174 | -4.4194 | -4.4194 | 7184 | -4.4194 | -4.4194 |
| 2018-05-01 12:30:00 | -6.4021 | -1.419 | 7174 | -6.4021 | -6.4021 | 7184 | -6.4021 | -6.4021 | 6732 | -6.4021 | -6.4021 |
| 2018-05-01 12:00:00 | -1.7971 | -1.0396 | 7174 | -1.7971 | -1.7971 | 7184 | -1.7971 | -1.7971 | 6732 | -1.7971 | -1.7971 |
| 2018-05-01 11:30:00 | 0.5276 | 0.2621 | 7174 | 0.5276 | 0.5276 | 7184 | 0.5276 | 0.5276 | 6732 | 0.5276 | 0.5276 |
| 2018-05-01 11:00:00 | -0.4941 | -0.1098 | 7174 | -0.4941 | -0.4941 | 6732 | -0.4941 | -0.4941 | 7184 | -0.4941 | -0.4941 |
| 2018-05-01 10:30:00 | -0.6783 | -0.1384 | 7184 | -0.6783 | -0.6783 | 6732 | -0.6783 | -0.6783 | 7174 | -0.6783 | -0.6783 |
| 2018-05-01 10:00:00 | -0.1293 | -0.0246 | 7184 | -0.1293 | -0.1293 | 6732 | -0.1293 | -0.1293 | 7174 | -0.1293 | -0.1293 |
| 2018-05-01 9:30:00 | -0.2703 | -0.0669 | 7184 | -0.2703 | -0.2703 | 6732 | -0.2703 | -0.2703 | 7174 | -0.2703 | -0.2703 |
+---------------------+-------------+-------------+------------------+------------------------------+------------------------------+--------------------+--------------------------------+--------------------------------+---------------------+---------------------------------+---------------------------------+
也许我们甚至可以更深入一级？并且在汇总时间戳时说，获取顶级组的顶级网址或顶级组网址组合？
很少有其他可能真正有用的聚合：
1）说出一个特定的时间范围，比如整整一个月： 由网址汇总，显示最佳/最差时间＆amp;整个范围的值，但也会在整个月的整个时间内对它们进行平均，并在那里获取聚合，如图所示：
+-----------+-------------+-------------+------------------------------------+-------------------------------------+--------------------------+----------------------------+--------------+----------------------------+----------------+------------------------------+
| url | aggregate_1 | aggregate_2 | best performance timestamp overall | worst performance timestamp overall | peak time of average day | trough time of average day | mean_at_peak | standard_deviation_at_peak | mean_at_trough | standard_deviation_at_trough |
+-----------+-------------+-------------+------------------------------------+-------------------------------------+--------------------------+----------------------------+--------------+----------------------------+----------------+------------------------------+
| abc.com | -0.3673 | -0.3673 | 2018-05-01 14:30:00 | 2018-05-01 14:30:00 | 2018-05-01 9:30:00 | 2018-05-01 9:30:00 | 0.5276 | 0.5276 | 0.5276 | 0.5276 |
| xyz.com | 4.2187 | 4.2187 | 2018-05-01 14:00:00 | 2018-05-01 14:00:00 | 2018-05-01 10:00:00 | 2018-05-01 10:00:00 | 0.5276 | 0.5276 | 0.5276 | 0.5276 |
| pqr.org | -2.3875 | -2.3875 | 2018-05-01 13:30:00 | 2018-05-01 13:30:00 | 2018-05-01 10:30:00 | 2018-05-01 10:30:00 | 4.2187 | 4.2187 | 4.2187 | 4.2187 |
| many.com | -4.4194 | -4.4194 | 2018-05-01 13:00:00 | 2018-05-01 13:00:00 | 2018-05-01 10:30:00 | 2018-05-01 10:30:00 | 5.449066667 | 5.449066667 | 5.449066667 | 5.449066667 |
| abc.com | -6.4021 | -6.4021 | 2018-05-01 12:30:00 | 2018-05-01 10:30:00 | 2018-05-01 12:00:00 | 2018-05-01 12:00:00 | 4.2187 | 4.2187 | 4.2187 | 4.2187 |
| xyz.com | -1.7971 | -1.7971 | 2018-05-01 12:00:00 | 2018-05-01 12:00:00 | 2018-05-01 10:30:00 | 2018-05-01 10:30:00 | 0.5276 | 0.5276 | 0.5276 | 0.5276 |
| pqr.org | 0.5276 | 0.5276 | 2018-05-01 11:30:00 | 2018-05-01 10:30:00 | 2018-05-01 10:30:00 | 2018-05-01 10:30:00 | 7.985716667 | 7.985716667 | 7.985716667 | 7.985716667 |
| many.com | -0.4941 | -0.4941 | 2018-05-01 11:00:00 | 2018-05-01 11:00:00 | 2018-05-01 11:00:00 | 2018-05-01 11:00:00 | 4.2187 | 4.2187 | 4.2187 | 4.2187 |
| many.com | -0.6783 | -0.6783 | 2018-05-01 10:30:00 | 2018-05-01 10:30:00 | 2018-05-01 9:30:00 | 2018-05-01 9:30:00 | 0.5276 | 0.5276 | 0.5276 | 0.5276 |
| sites.com | -0.1293 | -0.1293 | 2018-05-01 10:00:00 | 2018-05-01 10:00:00 | 2018-05-01 10:30:00 | 2018-05-01 10:30:00 | 9.522366667 | 9.522366667 | 9.522366667 | 9.522366667 |
| here.com | -0.2703 | -0.2703 | 2018-05-01 9:30:00 | 2018-05-01 9:30:00 | 2018-05-01 10:00:00 | 2018-05-01 10:00:00 | 4.2187 | 4.2187 | 4.2187 | 4.2187 |
+-----------+-------------+-------------+------------------------------------+-------------------------------------+--------------------------+----------------------------+--------------+----------------------------+----------------+------------------------------+
2）对于指定网址的列表或让查询本身构建网址列表，例如与每个窗口中使用metric_1相匹配的模式或前3个网址，显示所提供或所需指标的百分比贡献：
+---------------------+----------+-------------------------------+-------------------------------+-------------------------------+----------+-------------------------------+-------------------------------+-------------------------------+
| timestamp | metric_1 | contribution_percentage_url_1 | contribution_percentage_url_2 | contribution_percentage_url_3 | metric_2 | contribution_percentage_url_1 | contribution_percentage_url_2 | contribution_percentage_url_3 |
+---------------------+----------+-------------------------------+-------------------------------+-------------------------------+----------+-------------------------------+-------------------------------+-------------------------------+
| 2018-05-01 14:30:00 | -0.3673 | 33 | 26 | 18 | -0.3673 | 53 | 30 | 11 |
| 2018-05-01 14:00:00 | 4.2187 | 33 | 29 | 12 | 4.2187 | 30 | 32 | 20 |
| 2018-05-01 13:30:00 | -2.3875 | 53 | 29 | 17 | -2.3875 | 37 | 32 | 11 |
| 2018-05-01 13:00:00 | -4.4194 | 39 | 27 | 19 | -4.4194 | 31 | 34 | 10 |
| 2018-05-01 10:30:00 | -6.4021 | 41 | 25 | 15 | -6.4021 | 31 | 30 | 16 |
| 2018-05-01 12:00:00 | -1.7971 | 45 | 27 | 12 | -1.7971 | 32 | 30 | 12 |
| 2018-05-01 10:30:00 | 0.5276 | 50 | 35 | 18 | 0.5276 | 41 | 25 | 13 |
| 2018-05-01 11:00:00 | -0.4941 | 33 | 33 | 16 | -0.4941 | 44 | 34 | 13 |
| 2018-05-01 10:30:00 | -0.6783 | 53 | 33 | 18 | -0.6783 | 54 | 33 | 16 |
| 2018-05-01 10:00:00 | -0.1293 | 38 | 31 | 14 | -0.1293 | 42 | 31 | 17 |
| 2018-05-01 9:30:00 | -0.2703 | 30 | 35 | 11 | -0.2703 | 30 | 35 | 16 |
+---------------------+----------+-------------------------------+-------------------------------+-------------------------------+----------+-------------------------------+-------------------------------+-------------------------------+
3）透视： 对于提供的日期列表，或者从提供的日期开始的+ - 5天以及特定的关键度量：比较整个日期的度量：
+-------------+---------+-------------+-------------+-------------+-------------+---------+-------------+-------------+-------------+-------------+---------+
| time of day | date-5 | date-4 | date-3 | date-2 | date-1 | date | date+1 | date+2 | date+3 | date+4 | date+5 |
+-------------+---------+-------------+-------------+-------------+-------------+---------+-------------+-------------+-------------+-------------+---------+
| 14:30:00 | -0.3673 | 0.5276 | 0.5276 | 0.5276 | 0.5276 | -0.3673 | 0.5276 | 0.5276 | 0.5276 | 0.5276 | -0.3673 |
| 14:00:00 | 4.2187 | 0.5276 | 0.5276 | 0.5276 | 0.5276 | 4.2187 | 0.5276 | 0.5276 | 0.5276 | 0.5276 | 4.2187 |
| 13:30:00 | -2.3875 | 4.2187 | 4.2187 | 4.2187 | 4.2187 | -2.3875 | 4.2187 | 4.2187 | 4.2187 | 4.2187 | -2.3875 |
| 13:00:00 | -4.4194 | 5.449066667 | 5.449066667 | 5.449066667 | 5.449066667 | -4.4194 | 5.449066667 | 5.449066667 | 5.449066667 | 5.449066667 | -4.4194 |
| 12:30:00 | -6.4021 | 4.2187 | 4.2187 | 4.2187 | 4.2187 | -6.4021 | 4.2187 | 4.2187 | 4.2187 | 4.2187 | -6.4021 |
| 12:00:00 | -1.7971 | 0.5276 | 0.5276 | 0.5276 | 0.5276 | -1.7971 | 0.5276 | 0.5276 | 0.5276 | 0.5276 | -1.7971 |
| 11:30:00 | 0.5276 | 7.985716667 | 7.985716667 | 7.985716667 | 7.985716667 | 0.5276 | 7.985716667 | 7.985716667 | 7.985716667 | 7.985716667 | 0.5276 |
| 11:00:00 | -0.4941 | 4.2187 | 4.2187 | 4.2187 | 4.2187 | -0.4941 | 4.2187 | 4.2187 | 4.2187 | 4.2187 | -0.4941 |
| 10:30:00 | -0.6783 | 0.5276 | 0.5276 | 0.5276 | 0.5276 | -0.6783 | 0.5276 | 0.5276 | 0.5276 | 0.5276 | -0.6783 |
| 10:00:00 | -0.1293 | 9.522366667 | 9.522366667 | 9.522366667 | 9.522366667 | -0.1293 | 9.522366667 | 9.522366667 | 9.522366667 | 9.522366667 | -0.1293 |
| 9:30:00 | -0.2703 | 4.2187 | 4.2187 | 4.2187 | 4.2187 | -0.2703 | 4.2187 | 4.2187 | 4.2187 | 4.2187 | -0.2703 |
+-------------+---------+-------------+-------------+-------------+-------------+---------+-------------+-------------+-------------+-------------+---------+
4）有一个名为metric_lg的指标，它根据计数表示网址的生命周期或生命周期。因此，从指定日期或组的第一个时间戳开始，根据其计数计算某些度量聚合，即对于单个URL，范围将为1-5,5-10,10-20,20-50 ，50-80,80-200,200-1000,1000-10000,10000 +：让他们称他们为A，B，C，D，E，F，G，H，I阶段。然而，这个计数需要从小组开始的时候开始累积，即从小组中出现。假设一组7184是在2018-05-01 10:00:00启动的，而7174是在2018-04-30 12:00:00启动的，那么两组中出现的特定网址都会从其中累积其metric_lg相应组的开始，即7184中的生命周期阶段将是从7184开始的metric_lg累积，即2018-05-01 10:00:00，其生命周期阶段在7174中将是metric_lg的累积开始7174即2018-04-30 12:00:00。
因此，对于提供的组列表，这样的事情将有所帮助：根据metric_lg生命周期阶段计算其他度量聚合，并比较按生命周期阶段打破的组性能。
+---------------------+--------------------+---------------------+--------------------+--------------------+--------------------+--------------------+--------------------+--------------------+--------------------+
| timestamp | A_aggregate_metric | B _aggregate_metric | C_aggregate_metric | D_aggregate_metric | E_aggregate_metric | F_aggregate_metric | G_aggregate_metric | H_aggregate_metric | I_aggregate_metric |
+---------------------+--------------------+---------------------+--------------------+--------------------+--------------------+--------------------+--------------------+--------------------+--------------------+
| 2018-05-01 14:30:00 | -0.3673 | 0.5276 | 0.5276 | 0.5276 | 0.5276 | -0.3673 | 0.5276 | 0.5276 | 0.5276 |
| 2018-05-01 14:00:00 | 4.2187 | 0.5276 | 0.5276 | 0.5276 | 0.5276 | 4.2187 | 0.5276 | 0.5276 | 0.5276 |
| 2018-05-01 13:30:00 | -2.3875 | 4.2187 | 4.2187 | 4.2187 | 4.2187 | -2.3875 | 4.2187 | 4.2187 | 4.2187 |
| 2018-05-01 13:00:00 | -4.4194 | 5.449066667 | 5.449066667 | 5.449066667 | 5.449066667 | -4.4194 | 5.449066667 | 5.449066667 | 5.449066667 |
| 2018-05-01 10:30:00 | -6.4021 | 4.2187 | 4.2187 | 4.2187 | 4.2187 | -6.4021 | 4.2187 | 4.2187 | 4.2187 |
| 2018-05-01 12:00:00 | -1.7971 | 0.5276 | 0.5276 | 0.5276 | 0.5276 | -1.7971 | 0.5276 | 0.5276 | 0.5276 |
| 2018-05-01 10:30:00 | 0.5276 | 7.985716667 | 7.985716667 | 7.985716667 | 7.985716667 | 0.5276 | 7.985716667 | 7.985716667 | 7.985716667 |
| 2018-05-01 11:00:00 | -0.4941 | 4.2187 | 4.2187 | 4.2187 | 4.2187 | -0.4941 | 4.2187 | 4.2187 | 4.2187 |
| 2018-05-01 10:30:00 | -0.6783 | 0.5276 | 0.5276 | 0.5276 | 0.5276 | -0.6783 | 0.5276 | 0.5276 | 0.5276 |
| 2018-05-01 10:00:00 | -0.1293 | 9.522366667 | 9.522366667 | 9.522366667 | 9.522366667 | -0.1293 | 9.522366667 | 9.522366667 | 9.522366667 |
| 2018-05-01 9:30:00 | -0.2703 | 4.2187 | 4.2187 | 4.2187 | 4.2187 | -0.2703 | 4.2187 | 4.2187 | 4.2187 |
+---------------------+--------------------+---------------------+--------------------+--------------------+--------------------+--------------------+--------------------+--------------------+--------------------+
如果您需要数据上下文，请假设三个指标： metric_1：美元收入 metric_2：以美元计算的成本 metric_lg：数千的流量计数
PS：执行此操作是优于MySQL over python，因为其中一些将用于创建自定义VIEW，因此可以经常查看并进行进一步分析。这很多，非常感谢，真的