Question

我有一个包含四列的表格，如下所示：

id,name, key1, key2, key3, date
 1,'A' ,'x1','y1','z1','2015-11-11'
 2,'A' ,'x1','y2','z1','2015-11-11'
 3,'B' ,'x1','y2','z1','2015-11-11'
 4,'B' ,'x1','y1','z2','2015-11-11'
 5,'A' ,'x1','y1','z1','2015-11-12'
 6,'A' ,'x1','y2','z3','2015-11-12'
 7,'B' ,'x2','y3','z3','2015-11-12'
 8,'B' ,'x2','y1','z4','2015-11-12'
 9,'D' ,'x2','y1','z4','2015-11-12'

我想计算每个key1, key2, key3的不同新 date - s的数量：

2015-11-11  3                     -- (two distinct key1, key2, key3 row 2,3)
2015-11-12  3                     -- (new key and distinct key1, key2, key3)

我该怎么做？

Answer 1

假设您的钥匙没有|在他们里面的char（如果不是这样的话，选择另一个char：

SELECT DATE, EXACT_COUNT_DISTINCT(key) AS keys
FROM (
  SELECT DATE, key, LEAD(DATE) OVER(PARTITION BY key ORDER BY DATE DESC) AS new
  FROM (SELECT DATE, CONCAT(key1, '|', key2, '|', key3) AS key 
        FROM YourTable GROUP BY 1,2)
) WHERE new IS NULL
GROUP BY DATE 
ORDER BY DATE

您可以注意到，这只是我对您之前问题的回答。我将所有三个键合并为一个 - 并应用与以前完全相同的代码

为日期列和3键的每个值计算某列的不同值

1 个答案: