带有 case when 和 group by 的值列计数答案

【问题标题】：Count of Value column with case when and group by带有 case when 和 group by 的值列计数
【发布时间】：2020-09-06 09:48:32
【问题描述】：

表：

Date            Id       Value                    
2019-02-09      a1       ab            
2019-01-12      a2       cd
2019-12-12      a1       ab
2017-07-23      a1       ab
2018-12-09      a1       ab
2018-12-28      a1       cd
2016-11-09      a2       cd
2016-05-19      a2       ab

输出：

      Id        Max-Min               
      a1        1     
      a2       -1

目的是创建Max_year_count-Min_year_count per Id。
比如Value column中的计算：
(count of occurrence of value in max_year with group by Id)-(count of occurrence of value in min_year with group by Id)
谢谢！！

【问题讨论】：

标签： sql oracle oracle-sqldeveloper

【解决方案1】：

Oracle 支持 FIRST/LAST 聚合：

SELECT id,
    -- latest year's count
    Count(*) KEEP (Dense_Rank LAST  ORDER BY Extract(YEAR From "Date"))
     -- oldest year's count
  - Count(*) KEEP (Dense_Rank FIRST ORDER BY Extract(YEAR From "Date"))
FROM DATA
GROUP BY Id

【讨论】：

出于某种原因，第一个/最后一个永远不会留在我的脑海中......这绝对应该是公认的答案。

【解决方案2】：

您可以为每个Id 值创建一个最大和最小年份的派生表，然后计算与这些值匹配的id 值的年数并减去它们以获得Max-Min：

SELECT d.Id,
       COUNT(CASE WHEN EXTRACT(YEAR FROM "Date") = Max_Year THEN 1 END) -
       COUNT(CASE WHEN EXTRACT(YEAR FROM "Date") = Min_Year THEN 1 END) AS "Max-Min"
FROM data d
JOIN (SELECT Id, 
             MAX(EXTRACT(YEAR FROM "Date")) AS Max_Year,
             MIN(EXTRACT(YEAR FROM "Date")) AS Min_Year
      FROM data
      GROUP BY Id) m ON m.Id = d.Id
GROUP BY d.Id

输出：

ID  Max-Min
a1  1
a2  -1

Demo on dbfiddle

【讨论】：