Question

我在数据库中有一个证券价格数据集。数据的结构如下：

id      security_id     time_to_maturity      price
001         01               1.5               100.45
002         01               1.3               101.45
003         01               1.1               102.45
004         01               1.02              101.45
005         01               1.0               101.45
006         03              22.3               94.45
007         03              22.1               96.45
008         03              21.8               98.45
009         05               4.2               111.45
010         05               4.1               112.45
011         05               3.8               111.45
...

id是row_id，security_id是每个安全的ID。我试图获取每个安全性的特定时间范围内的数据。首先，我运行一个查询来查找每个安全ID的最小值和最大值，然后找到最小值和最大值之间的差值，最后找到一个比最小值大10％的值，如下所示：

SELECT security_id, MIN(time_to_maturity), MAX(time_to_maturity),
    MAX(time_to_maturity) - MIN(time_to_maturity) tDiff,
    ((MAX(time_to_maturity) - MIN(time_to_maturity)) * .1) + MIN(time_to_maturity)
  FROM db1
  group by security_id
  order by security_id

这给了我以下内容：

security_id    min()     max()     diff      min+(diff*.1)
  01             1.0        1.5      .5            1.05
  03            21.8       22.3      .5           21.85
  05             3.8        4.2      .4            3.84

最后，我要做的是从主数据集中仅选择security_id中每个time_to_maturity is < min+(diff*.1)的行。

我不知道如何构建它，因为我觉得我需要一个循环来通过security_id对数据进行子集化，然后是time_to_maturity is < min+(diff*.1)。

答案看起来像这样：

id      security_id     time_to_maturity      price
004         01               1.02              101.45
005         01               1.0               101.45
008         03              21.8               98.45
011         05               3.8               111.45

有什么建议吗？

Answer 1

SELECT A.id,B.security_id,A.time_to_maturity,A.price
FROM db1 A,
(
SELECT security_id, MIN(time_to_maturity) AS min_time_to_maturity, MAX(time_to_maturity) AS max_time_to_maturity,
    MAX(time_to_maturity) - MIN(time_to_maturity) tDiff,
    ((MAX(time_to_maturity) - MIN(time_to_maturity)) * .1) + MIN(time_to_maturity)
  FROM db1
  group by security_id
  order by security_id
) B
WHERE A.security_id = B.security_id
  AND A.time_to_maturity < (B.min_time_to_maturity+(B.tdiff*0.1));

PS：这只适用于MYSQL。

Answer 2

您没有说出您使用的是哪个版本的SQL Server，但假设它是2005+，您可以使用公用表表达式：

with cte as ( 
    SELECT security_id, 
        ((MAX(time_to_maturity) - MIN(time_to_maturity)) * .1) + MIN(time_to_maturity) as threshold
    FROM db1
    group by security_id
)
select id, db1.security_id, time_to_maturity, price
from db1
inner join cte
   on db1.security_id = cte.security_id
where time_to_maturity < threshold

SQL匹配基于2个条件的行

2 个答案: