Tsql Windows函数中的除以0错误

我试图使用Windows函数计算每月已结案索赔的百分比除以总索赔。但某些月份的索赔总额为零，这将导致出现

除以零的错误消息

我试着写一个案例陈述来处理索赔总额等于零的情况，将百分比设置为零，但到目前为止我没有任何运气

下面是我期待的屏幕截图：


下面的TSQL是我在错误处理被零除的消息时所做的尝试：我在除法器的底部添加了一个null if，并在一个CASE语句中尝试在底部除法器为零时设置值=0，这两个操作都会导致相同的错误。有没有关于如何避免这个错误的建议

CREATE TABLE #ClaimCounts
(
Year INT,
ClaimStatus VARCHAR (50),
LossMonth DATE,
ClaimMonth DATE,
ClaimCount INT
);

INSERT INTO #ClaimCounts
(
Year,
ClaimStatus,
LossMonth,
ClaimMonth,
ClaimCount
)
VALUES
(2008, 'Closed', '20080630', '20080131', 0),
(2008, 'Total', '20080630', '20080131', 0),
(2008, 'Closed', '20080630', '20080229', 0),
(2008, 'Total', '20080630', '20080229', 0),
(2008, 'Closed', '20080630', '20080331', 0),
(2008, 'Total', '20080630', '20080331', 0),
(2008, 'Closed', '20080630', '20080430', 0),
(2008, 'Total', '20080630', '20080430', 0),
(2008, 'Closed', '20080630', '20080531', 0),
(2008, 'Total', '20080630', '20080531', 0),
(2008, 'Closed', '20080630', '20080630', 0),
(2008, 'Total', '20080630', '20080630', 6),
(2008, 'Closed', '20080630', '20090731', 2),
(2008, 'Total', '20080630', '20090731', 5),
(2008, 'Closed', '20080630', '20080831', 1),
(2008, 'Total', '20080630', '20080831', 1),
(2008, 'Closed', '200806308', '20080930', 3),
(2008, 'Total', '20080630', '20080930', 3),
(2008, 'Closed', '20080630', '20081031', 2),
(2008, 'Total', '20080630', '20081031', 3),
(2008, 'Closed', '200806308', '20081130', 0),
(2008, 'Total', '20080630', '20081130', 0);

SELECT Year,
   ClaimStatus,
   LossMonth,
   ClaimMonth,
   ClaimCount,
   SUM ( CASE WHEN ClaimStatus = 'Closed' THEN ISNULL ( ClaimCount, 0 ) * 1.0 ELSE 0 END ) OVER (PARTITION BY Year, ClaimMonth) / 
   SUM ( CASE WHEN ClaimStatus = 'Total' THEN ClaimCount * 1.0 ELSE 0 END ) OVER (PARTITION BY Year, ClaimMonth) AS PercentageClosedClaims1 ,

   SUM ( CASE WHEN ClaimStatus = 'Closed' THEN ISNULL ( ClaimCount, 0 ) * 1.0 ELSE 0 END ) OVER (PARTITION BY Year, ClaimMonth) / 
   SUM ( CASE WHEN ClaimStatus = 'Total' THEN NULLIF(ClaimCount, 0) * 1.0 ELSE 0 END ) OVER (PARTITION BY Year, ClaimMonth) AS PercentageClosedClaims2 ,

   CASE WHEN ClaimStatus = 'Total' AND ClaimCount = 0 THEN 0
       ELSE SUM ( CASE WHEN ClaimStatus = 'Closed' THEN ISNULL ( ClaimCount, 0 ) * 1.0 ELSE 0 END ) OVER (PARTITION BY Year, ClaimMonth) / 
            SUM ( CASE WHEN ClaimStatus = 'Total' THEN ISNULL ( ClaimCount, 0 ) * 1.0 ELSE 0 END ) OVER (PARTITION BY Year, ClaimMonth)
   END AS PercentageClosedClaims3
FROM #ClaimCounts;

DROP TABLE IF EXISTS #ClaimCounts;

这里有一个可行的选择。
IsNull（）
是可选的。我个人不介意
NULL
值

SELECT Year
      ,ClaimStatus
      ,LossMonth
      ,ClaimMonth
      ,ClaimCount
      ,PercentageClosedClaims1 = IsNull(SUM ( CASE WHEN ClaimStatus = 'Closed' THEN ISNULL ( ClaimCount, 0 ) * 1.0 ELSE 0 END ) OVER (PARTITION BY Year, ClaimMonth) 
                                        / nullif(SUM ( CASE WHEN ClaimStatus = 'Total' THEN ClaimCount * 1.0 ELSE 0 END ) OVER (PARTITION BY Year, ClaimMonth),0) 
                                        ,0)
      ,PercentageClosedClaims2 = IsNull(SUM ( CASE WHEN ClaimStatus = 'Closed' THEN ISNULL ( ClaimCount, 0 ) * 1.0 ELSE 0 END ) OVER (PARTITION BY Year, ClaimMonth) 
                                        / nullif(SUM ( CASE WHEN ClaimStatus = 'Total' THEN NULLIF(ClaimCount, 0) * 1.0 ELSE 0 END ) OVER (PARTITION BY Year, ClaimMonth),0) 
                                        ,0)
      ,PercentageClosedClaims3 = IsNull(SUM ( CASE WHEN ClaimStatus = 'Closed' THEN ISNULL ( ClaimCount, 0 ) * 1.0 ELSE 0 END ) OVER (PARTITION BY Year, ClaimMonth) 
                                        / nullif(SUM ( CASE WHEN ClaimStatus = 'Total' THEN ISNULL ( ClaimCount, 0 ) * 1.0 ELSE 0 END ) OVER (PARTITION BY Year, ClaimMonth),0)
                                        ,0)

FROM #ClaimCounts;

IsNull（分子/NullIf（分母，0），0）DDL和DML很好，但请尽量避免使用特定语言的日期。对于世界上大多数国家来说，一年中没有30个月。：）ISO格式
yyyyMMdd
和
yyyy-MM-ddThh:MM:ss.nnnnnnn
始终是首选格式，因为这样人们就不需要更改语言设置或修改您的数据，使其易于使用。谢谢。谢谢@Larnu-我已经更新了TSQL。@johncapelletti-我不确定如何将您的答案应用到我的代码中。@MISNole请参阅下面的答案谢谢John！为了提醒我分子和分母这两个术语，我完全忘记了这些术语。