SQL：查找按用户分组的最新两个记录之间的差异

Posted 2021-04-27

假设我有一个包含三列的表格：[user_id，created_at，文本]

让U为至少具有两个记录的用户集。如何找到最近两个记录中的文本之间没有差异的U百分比？

答案

您可以使用窗口函数和聚合：

select user_id,
       avg(case when min(text) = max(text) then 1.0 else 0 end) as ratio_same
from (select t.*,
             row_number() over (partition by user_id order by created_at desc) as seqnum
      from t
     ) t
where seqnum <= 2
group by user_id
having max(seqnum = 2)   -- make sure there are two records

另一答案

只是一个记事本涂鸦（未经测试）

SELECT 100.0*SUM(samePrevText)/COUNT(*) as Perc
FROM
(
    SELECT user_id, created_at, text,
     row_number() over (partition by user_id order by created_at desc) as rn,
     case when text = lead(text) over (partition by user_id order by created_at desc) then 1 else 0 end as samePrevText
    FROM usertexts
) q
WHERE rn = 1

另一答案

虽然不漂亮，但是似乎可以解决问题：

SELECT SUM((LatestTwoRowsEqual) * 1.00) / COUNT(DISTINCT user_id) AS UsersPercentage
FROM (
  SELECT user_id,
      CASE 
        WHEN 
          ROW_NUMBER() OVER(
            PARTITION BY user_id 
            ORDER BY created_at DESC
          ) <= 2 AND -- Only look at two latest rows per user_id
          MAX(text) OVER(
            PARTITION BY user_id 
            ORDER BY created_at DESC 
            ROWS BETWEEN 1 PRECEDING AND 1 PRECEDING) = text -- Check if values are the same
        THEN 1
        ELSE 0
      END LatestTwoRowsEqual
  FROM MyTable
  WHERE user_id IN ( -- Only get users with at least two records
    SELECT user_id 
    FROM MyTable 
    GROUP BY user_id
    HAVING COUNT(*) > 1 
  )
) src

如果您的DBMS支持LAG功能，也可以使用它。

SQL Fiddle