sql - 如何确定 SQL 表(雪花)中出现次数最多的值并考虑关系？

Question

假设我们有如下数据:

ID    tag     data            timestamp
001     A   walter  2021-06-04 09:46:25
005     F   junior  2021-06-05 09:47:25
001     B   junior  2021-06-04 09:47:25
002     C  soprano  2021-06-04 09:48:25
002     C     alto  2021-06-04 09:49:25
001     A    brown  2021-06-04 09:50:25
003     A   cleave  2021-06-04 09:51:25
003     B     land  2021-06-04 09:52:25
004     C   before  2021-06-04 09:53:25
005     H   junior  2021-06-04 09:47:25

我需要知道每个 ID 值中出现次数最多的 tag。如果出现平局，请使用最新 ID 标签，以时间戳表示。

预期结果:

ID   tag
001    A
002    C
003    B
004    C
005    F

Answer 1

使用QUALIFY和RANK过滤分组结果:

SELECT ID, tag, COUNT(*) AS cnt, MAX(timestamp) AS max_t
FROM tab
GROUP BY ID, tag
QUALIFY RANK() OVER(PARTITION BY ID ORDER BY cnt DESC, max_t DESC) = 1

---

示例数据:

CREATE OR REPLACE TABLE tab(ID STRING, tag STRING, data  STRING, timestamp TIMESTAMP)
AS
          SELECT '001',     'A' ,' walter','2021-06-04 09:46:25'
UNION ALL SELECT '005',     'F' ,' junior','2021-06-05 09:47:25'
UNION ALL SELECT '001',     'B' ,' junior','2021-06-04 09:47:25'
UNION ALL SELECT '002',     'C' ,'soprano','2021-06-04 09:48:25'
UNION ALL SELECT '002',     'C' ,'   alto','2021-06-04 09:49:25'
UNION ALL SELECT '001',     'A' ,'  brown','2021-06-04 09:50:25'
UNION ALL SELECT '003',     'A' ,' cleave','2021-06-04 09:51:25'
UNION ALL SELECT '003',     'B' ,'   land','2021-06-04 09:52:25'
UNION ALL SELECT '004',     'C' ,' before','2021-06-04 09:53:25'
UNION ALL SELECT '005',     'H' ,' junior','2021-06-04 09:47:25';

简化查询:

SELECT ID, tag
FROM tab
GROUP BY ID, tag
QUALIFY RANK() OVER(PARTITION BY ID ORDER BY COUNT(*) DESC, MAX(timestamp) DESC) = 1
ORDER BY ID;

输出: