mysql滑动聚合/年初至今聚合原理与用法实例分析

这篇文章主要介绍了mysql滑动聚合原理与用法,结合实例形式分析了mysql滑动聚合的相关功能、原理、使用方法及操作注意事项,需要的朋友可以参考下

本文实例讲述了mysql滑动聚合/年初至今聚合原理与用法。分享给大家供大家参考，具体如下：

滑动聚合是按顺序对滑动窗口范围内的数据进行聚合的操作。下累积聚合不同，滑动聚合并不是统计开始计算的位置到当前位置的数据。

这里以统计最近三个月中员工第月订单情况为例来介绍滑动聚合。

滑动聚合和累积聚合解决方案的主要区别在于连接的条件不同。滑动聚合条件不再是b.ordermonth <= a.ordermonth,而应该是b.ordermonth大于前三个月的月份，并且小于当前月份。因此滑动聚合的解决方案的SQL语句如下

 SELECT a.empid, DATE_FORMAT(a.ordermonth, '%Y-%m') AS ordermonth, a.qty AS thismonth, SUM(b.qty) AS total, CAST(AVG(b.qty) AS DECIMAL(5,2)) AS avg FROM emporders a INNER JOIN emporders b ON a.empid=b.empid AND b.ordermonth > DATE_ADD(a.ordermonth, INTERVAL -3 MONTH) AND b.ordermonth <= a.ordermonth WHERE DATE_FORMAT(a.ordermonth,'%Y')='2015' AND DATE_FORMAT(b.ordermonth,'%Y')='2015' GROUP BY a.empid,DATE_FORMAT(a.ordermonth, '%Y-%m'),a.qty ORDER BY a.empid,a.ordermonth

运行结果如下

该解决方案返回的是三个月为一个周期的滑动聚合，但是每个用户包含前两个月并且未满3个月的聚合。如果只希望返回满3个月的聚合，不返回未满3个月的聚合，可以使用HAVING过滤器进行过滤，过滤的条件为MIN(b.ordermonth)=DATE_ADD(a.ordermonth, INTERVAL -2 MONTH),例如

 SELECT a.empid, a.ordermonth AS ordermonth, a.qty AS thismonth, SUM(b.qty) AS total, CAST(AVG(b.qty) AS DECIMAL(5,2)) AS avg FROM emporders a INNER JOIN emporders b ON a.empid=b.empid AND b.ordermonth > DATE_ADD(a.ordermonth, INTERVAL -3 MONTH) AND b.ordermonth <= a.ordermonth WHERE DATE_FORMAT(a.ordermonth,'%Y')='2015' AND DATE_FORMAT(b.ordermonth,'%Y')='2015' AND a.empid=1 GROUP BY a.empid,DATE_FORMAT(a.ordermonth, '%Y-%m'),a.qty HAVING MIN(b.ordermonth)=DATE_ADD(a.ordermonth, INTERVAL-2 MONTH) ORDER BY a.empid,a.ordermonth

运行结果如下

年初至今聚合和滑动聚合类似，不同的地方仅在于统计的仅为当前一年的聚合。唯一的区别体现在下限的开始位置上。在年初至今的问题中，下限为该年的第一天，而滑动聚合的下限为N个月的第一天。因此，年初至今的问题的解决方案如下图所示，得到的结果

 SELECT a.empid, DATE_FORMAT(a.ordermonth, '%Y-%m') AS ordermonth, a.qty AS thismonth, SUM(b.qty) AS total, CAST(AVG(b.qty) AS DECIMAL(5,2)) AS avg FROM emporders a INNER JOIN emporders b ON a.empid=b.empid AND b.ordermonth >= DATE_FORMAT(a.ordermonth, '%Y-01-01') AND b.ordermonth <= a.ordermonth AND DATE_FORMAT(b.ordermonth,'%Y')='2015' GROUP BY a.empid,a.ordermonth,a.qty ORDER BY a.empid,a.ordermonth

运行结果如下