sql计算2行之间的差异

I'm looking for a methodology to compare the difference between 2 rows in the same table. From what I found here (How to get difference between two rows for a column field?) it's almost what I wanted. I have done the following code:

我正在寻找一种方法来比较同一个表中2行之间的差异。从我在这里找到的(如何获得列字段的两行之间的差异?)这几乎是我想要的。我做了以下代码:

create table #tmpTest
(
    id_fund int null,
    id_ShareType int null,
    ValueDate datetime null,
    VarNAV float null,
    FundPerf float null,
)

insert into #tmpTest(id_fund, id_ShareType, ValueDate, VarNAV)
values(1,1,'20140101',100)
insert into #tmpTest(id_fund, id_ShareType, ValueDate, VarNAV) 
values(1,1,'20140102',20)

update #tmpTest
set hrc.FundPerf = (isnull(hrn.VarNAV, 0) - hrc.VarNAV)/hrc.VarNAV
from #tmpTest hrc 
left join #tmpTest hrn on hrn.ValueDate = (select min(ValueDate) from #tmpTest where ValueDate > hrc.ValueDate)
and hrc.id_fund = hrn.id_fund and hrc.id_ShareType = hrn.id_ShareType

My issue is that the result I'm computing starts on line 1 instead of line 2.

我的问题是,我正在计算的结果从第1行而不是第2行开始。

Hereunder the result I'm obtaining:

下面是我得到的结果:

id_fund id_ShareType ValueDate           VarNAV                       FundPerf                     
------- ------------ ------------------- ------- -----------------------------
      1            1 2014-01-01 00:00:00     100                          -0.8
      1            1 2014-01-02 00:00:00      20                            -1

whereas I'd like it to be that way:

而我希望它是这样的:

id_fund id_ShareType ValueDate           VarNAV                       FundPerf                     
------- ------------ ------------------- ------- -----------------------------
      1            1 2014-01-01 00:00:00     100                            -1
      1            1 2014-01-02 00:00:00      20                          -0.8

What's wrong with my approach?

我的方法有什么问题?

3 个解决方案

#1

You are not restricting the minimum to the same fund and share type.

您不是将最低限制为同一基金和股票类型。

update #tmpTest
    set hrc.FundPerf = (isnull(hrn.VarNAV, 0) - hrc.VarNAV)/hrc.VarNAV
    from #tmpTest hrc left join
         #tmpTest hrn
         on hrn.ValueDate = (select min(ValueDate)
                             from #tmpTest tt
                             where tt.ValueDate > hrc.ValueDate and
                                   hrc.id_fund = tt.id_fund and hrc.id_ShareType = tt.id_ShareType 
                            ) and
           hrc.id_fund = hrn.id_fund and hrc.id_ShareType = hrn.id_ShareType ;

#2

Try this:

update hrn
set FundPerf = (isnull(hrn.VarNAV, 0) - hrc.VarNAV)/hrc.VarNAV
from #tmpTest hrc 
left join #tmpTest hrn on hrn.ValueDate = (select min(ValueDate) from #tmpTest where ValueDate > hrc.ValueDate)
and hrc.id_fund = hrn.id_fund and hrc.id_ShareType = hrn.id_ShareType

#3

Hi you can achieve this using by CTE (Common Table Expression)

嗨,你可以通过CTE(公用表表达式)实现这一点

create table #tmpTest
(
    id_fund int null,
    id_ShareType int null,
    ValueDate datetime null,
    VarNAV float null,
    FundPerf float null,
)

insert into #tmpTest(id_fund, id_ShareType, ValueDate, VarNAV)
values(1,1,'20140101',100)
insert into #tmpTest(id_fund, id_ShareType, ValueDate, VarNAV) 
values(1,1,'20140102',20)

;With tbl as

( Select Row_Number() OVER (Order by T.ValueDate) as RowNumber,* From #tmpTest T )SELECT Cur.*,(ISNULL(Cur.VarNAV,0) - ISNULL(Prv.VarNAV,0))/Prv.VarNAV as [Col Name] FROM tbl Cur LEFT OUTER JOIN tbl Prv ON Cur.RowNumber = Prv.RowNumber+1 ORDER BY Cur.ValueDate

(选择Row_Number()OVER(按T.ValueDate排序)为RowNumber,*来自#tmpTest T)SELECT Cur。*,(ISNULL(Cur.VarNAV,0) - ISNULL(Prv.VarNAV,0))/ Prv.VarNAV as [Col Name] FROM tbl Cur LEFT OUTER JOIN tbl Prv ON Cur.RowNumber = Prv.RowNumber + 1 ORDER BY Cur.ValueDate

#1