Question

我想给CardId和InvoiceNo的每个组合提供以下数据的数字/等级

;WITH Test (CardId,InvoiceNo,Category,Amount) as 
(
    SELECT '1001','3001','Fees',30
    UNION ALL   SELECT '1001','3001','Non-Fees',45
    UNION ALL   SELECT '1001','3001','Service Fees',55
    UNION ALL   SELECT '1002','3002','Fees',10
    UNION ALL   SELECT '1002','3002','Non-Fees',25
    UNION ALL   SELECT '1002','3002','Service Fees',10
    UNION ALL   SELECT '1003','3010','Fees',45
    UNION ALL   SELECT '1003','3010','Non-Fees',70


)

喜欢这个

CardId  InvoiceNo   Amount  Rank
1001    3001         30     1
1001    3001         45     1
1001    3001         55     1
1002    3002         10     2
1002    3002         25     2
1002    3002         10     2
1003    3010         45     3
1003    3010         70     3

我使用Row_number()和Rank()尝试了以下查询，但未给出我想要的结果。 rank()将所有行排名为1，row_number()将每个组编号为1,2,3。

SELECT CardId
,InvoiceNo
,Amount
,RANK() OVER (PARTITION BY CardID,InvoiceNo ORDER BY CardId) as RankNo

FROM Test

Answer 1

删除分区。 rank函数在分区中查找分区，其中新分区重置排名。因此，它看到每个分区中的每条记录都与第一条相关。此外，您可能需要dense_rank()。

SELECT CardId ,InvoiceNo ,Amount ,DENSE_RANK() OVER (ORDER BY CardId, InvoiceNo) as RankNo FROM Test ORDER BY CardId, InvoiceNo, Amount

小提琴：

http://sqlfiddle.com/#!18/0f0c2/7

Answer 2

试试这个。我认为这就是你要找的东西。

SELECT CardId
    ,InvoiceNo
    ,Amount
    ,DENSE_RANK() OVER (ORDER BY CardId, InvoiceNo) as RankNo

FROM Test

如何在T-SQL中对分区进行排序

2 个答案: