双倍增量，其中第二个增量反映了 sql 中遇到数据的第一个

Question

我正在构建 healthcare 837 encounters，需要在 HL 段上设置增量。

C1 基于 Criteria1 的设置，C2 基于 Criteria2.
C2 永远不会与 C1 有相同的数字，反之亦然。
C1 我可以使用 row_number() over(order by (select Criteria1))

这是 C2 我遇到的问题。

 C1 | C2 | Criteria1 | Criteria2
  1 | 2 | ID1 | NID1
  1 | 3 | ID1 | NID2
  1 | 4 | ID1 | NID3
  5 | 6 | ID2 | NID4
  5 | 7 | ID2 | NID5
  5 | 8 | ID2 | NID6
  9 |10 | ID3 | NID7

简化查询：

 SELECT cm.Criteria1, cm.Criteria2, cj.C1 
 FROM [dbo].[TBL1] cm 
 JOIN (
         SELECT cm.Criteria1,
                row_number() over(order by (select Criteria1)) as C1 
         FROM [dbo].[TBL1] cm 
         GROUP BY cm.Criteria1) cj on cj.Criteria1 = cm.Criteria1 
 GROUP BY cm.Criteria1, cm.Criteria2, cj.C1 Order by cj.C1

Answer 1

不太清楚你到底想得到什么，但看起来下面就是你要找的东西：

with counts as (select count(distinct cm.criteria1) c1, count(distinct cm.criteria2) c2 from dbo.tbl1) 
select cj1.c1, cj2.c2, cm.criteria1, cm.criteria2
from dbo.tbl1 cm
inner join ( 
      select cm1.criteria1, 
         row_number() over ( order by cm1.criteria1)as c1 
      from dbo.tbl1 cm1 group by cm1.criteria1) cj1 
              on cj1.criteria1 = cm.criteria1
inner join ( 
      select cm2.criteria2, 
         (select counts.c1 from counts) + row_number() over ( order by cm2.criteria2) as c2 
      from dbo.tbl1 cm2 group by cm1.criteria2) cj2 
              on cj2.criteria2 = cm.criteria2
group by cm.criteria1, cm.criteria2, cj1.c1, cj2.c2
order by cj1.c1, cj2.c2

Answer 2

这似乎可行，但我没有检查很多边缘情况（开窗很有趣！）：

with tbl1 as (
    select 'ID1' as Criteria1, 'NID1' as Criteria2
    union
    select 'ID1', 'NID2'
    union
    select 'ID2', 'NID4'
    union
    select 'ID2', 'NID5'
    union
    select 'ID3', 'NID7'
)

select 
    rank() over (order by Criteria1) + DENSE_ranK() OVER (ORDER BY CRITERIA1) - 1 as C1,
    rank() over (order by Criteria1) + row_number() over (partition by Criteria1 order by Criteria2) + DENSE_ranK() OVER (ORDER BY CRITERIA1) - 1 as C2,
    Criteria1,
    Criteria2
from
  tbl1

稍微分解一下：

按照SQL 的说法，我们将每组 Criteria1 行称为 "partition"。

因此要求是：

C1 始终等于所有先前分区中的行数 + 当前分区的 1，加上先前分区的数量。

C2总是等于之前所有分区的行数+当前分区的1，加上之前的分区数，加上分区内所有之前的行数+当前行的1 .

RANK() over (order by Criteria1) 为您提供所有先前分区中的行数 + 1。

DENSE_RANK() over (order by Criteria1) - 1 给出之前分区的数量。

ROW_NUMBER() over (partition by Criteria1 order by Criteria2) 为您提供分区中之前的行数。

双倍增量，其中第二个增量反映了 sql 中遇到数据的第一个

Double increment where 2nd increment reflects 1st in sql for encounter data

sql

auto-increment