使用Count with Case表达式

Question

我可能会因为询问而分崩离析，但为什么在比较两种不同场景中的Claims_Completed时会产生不同的结果呢？ Claims_Completed值与两种情况下的Claims_Received相同。

SELECT
    DischargeType
   ,COUNT(CASE WHEN (DateReceived > '2/1/2015' AND DateReceived < DATEADD(dd, 1, '2/28/2015')) THEN 1 ELSE 0 END) AS Claims_Received
   ,COUNT(CASE WHEN (DateCompleted > '2/1/2015' AND DateCompleted < DATEADD(dd, 1, '2/28/2015')) THEN 1 ELSE 0 END) AS Claims_Completed 
FROM Claims GROUP BY DischargeType

场景2：

SELECT COUNT(*) AS Claims_Received 
FROM Claims 
WHERE DateReceived > '2/1/2015' AND DateReceived < DATEADD(dd, 1, '2/28/2015') 
GROUP BY DischargeType

SELECT COUNT(*) AS Claims_Completed 
FROM Claims 
WHERE DateCompleted > '2/1/2015' AND DateCompleted < DATEADD(dd, 1, '2/28/2015') 
GROUP BY DischargeType

Answer 1

你需要sum()而不是count() ：

SELECT DischargeType, 
      SUM(CASE WHEN (DateReceived > '2/1/2015' AND
                     DateReceived < DATEADD(dd, 1, '2/28/2015')) 
               THEN 1 ELSE 0 
          END) AS Claims_Received,
      SUM(CASE WHEN (DateCompleted > '2/1/2015' AND 
                     DateCompleted < DATEADD(dd, 1, '2/28/2015')) 
               THEN 1 ELSE 0 
          END) AS Claims_Completed 
FROM Claims 
GROUP BY DischargeType;

因为count()在这里误解，因为它会将0作为值或数据计数。

Answer 2

COUNT将计算序列中的所有非空值。 由于1和0都是非null，因此它计算所有内容。 一种选择是使用NULL而不是0：

SELECT
    DischargeType
   ,COUNT(CASE WHEN (DateReceived > '2/1/2015' 
               AND DateReceived < DATEADD(dd, 1, '2/28/2015')) 
               THEN 1 ELSE NULL END) AS Claims_Received
   ,COUNT(CASE WHEN (DateCompleted > '2/1/2015' 
               AND DateCompleted < DATEADD(dd, 1, '2/28/2015')) 
               THEN 1 ELSE NULL END) AS Claims_Completed 
FROM Claims 
GROUP BY DischargeType