SQL查询/计算基于日期和ID的值的排名

Question

对于每个日期和 ID，我们必须按降序排列值。

下面是输入表。

>input data
+------------+----+-------+
|    date    | id | value |
+------------+----+-------+
| 01-01-2018 | A  |    20 |
| 01-01-2018 | A  |    50 |
| 01-01-2018 | C  |    40 |
| 01-01-2018 | B  |    40 |
| 02-01-2018 | A  |    30 |
| 03-01-2018 | C  |    20 |
| 03-01-2018 | C  |    40 |
| 04-01-2018 | B  |     0 |
| 04-01-2018 | B  |    40 |
| 05-01-2018 | B  |    70 |
+------------+----+-------+

输出应如下所示：

>output data
+------------+----+-------+------+
|    date    | id | value | rank |
+------------+----+-------+------+
| 01-01-2018 | A  |    50 |    1 |
| 01-01-2018 | A  |    20 |    2 |
| 01-01-2018 | B  |    40 |    1 |
| 01-01-2018 | C  |    40 |    1 |
| 02-01-2018 | A  |    30 |    1 |
| 03-01-2018 | C  |    40 |    1 |
| 03-01-2018 | C  |    20 |    2 |
| 04-01-2018 | B  |    40 |    1 |
| 04-01-2018 | B  |     0 |    2 |
| 05-01-2018 | B  |    70 |    1 |
+------------+----+-------+------+

Answer 1

您可以使用RANK() ，按date和id进行分区并按value降序排序：

SELECT *,
       RANK() OVER (PARTITION BY date, id ORDER BY value DESC) AS ranking
FROM data

输出：

date        id  value   ranking
01-01-2018  A   50      1
01-01-2018  A   20      2
01-01-2018  B   40      1
01-01-2018  C   40      1
02-01-2018  A   30      1
03-01-2018  C   40      1
03-01-2018  C   20      2
04-01-2018  B   40      1
04-01-2018  B   0       2
05-01-2018  B   70      1

SQLFiddle 上的演示

此查询将在您标记问题的所有 DBMS 上运行。