SQL给定多个阈值，获取低于每个阈值的最大值

Question

我有多个阈值(通常是 2 个，但可能会有所不同)。我想为每个阈值找到每个类别中最大值小于或等于阈值的行。

例如给定类别 2 和阈值 5 和 10:

mytable:
category | val | data
---------+-----+---------
1        | 1   | 'foo'
1        | 3   | 'bar'
1        | 4   | 'baz'
2        | 2   | 'quz'
2        | 5   | 'wibble'
2        | 6   | 'wobble'
2        | 8   | 'ham'
2        | 12  | 'spam'
3        | 1   | 'eggs'

所以结果应该是:

category | val | data
---------+-----+---------
1        | 4   | 'baz'     \
2        | 5   | 'wibble'  | These are <= threshold 5
3        | 1   | 'eggs'    /
1        | 4   | 'baz'     \
2        | 8   | 'ham'     | These are <= threshold 10
3        | 1   | 'eggs'    /

注意:如果行不同也可以，但不是必需的。

到目前为止，我只能对 1 个阈值进行查询(基本上是标准的每组最大 n 次查询):

SELECT t1.category, t1.val, t1.data
FROM mytable t1
JOIN (
  SELECT category, MAX(val) AS val
  FROM mytable
  GROUP BY category
  WHERE val < @threshold
) AS t2
ON t1.category=t2.category AND t1.val=t2.val

如何处理多个阈值？

如果重要的话，我使用的是 T-SQL。通用的 SQL 查询会很好，但不是必需的。

Answer 1

我只需将阈值指定为行，以便更容易加入。

DECLARE @t TABLE (category int, val int, data varchar(10));
INSERT INTO @t VALUES
(1, 1, 'foo'),
(1, 3, 'bar'),
(1, 4, 'baz'),
(2, 2, 'quz'),
(2, 5, 'wibble'),
(2, 6, 'wobble'),
(2, 8, 'ham'),
(2, 12, 'spam'),
(3, 1, 'eggs');

SELECT *
FROM (
    SELECT *, ROW_NUMBER() OVER (PARTITION BY threshold, category ORDER BY val DESC) AS rn
    FROM (VALUES
        (5),
        (10)
    ) thresholds(threshold)
    JOIN @t AS t ON val <= threshold
) AS x
WHERE rn = 1
ORDER BY threshold, category

如果(VAULES ...)子句不可用，您可以简单地使用FROM (SELECT 5 AS Threshold UNION ALL SELECT 10)阈值。