Question

有一种很好的方法可以让一个查询将值乘以相同计算值（本身）与上一行按某一列分组。下面我有一个表格，其中有两个按日期排序的帐户。（轻微扭曲，如果没有前一行使用1）。值是计算值，并分别在每个帐户上计算。

注意：我已在每个查询结果的括号中显示了Value的计算。

初始表

 Account      Date             ChangePct       
 547          10/24/17           .997715     
 547          10/26/17           .973493     
 547          10/27/17          1.001300     
 547          10/30/17           .999541     
 600          10/26/17          1.004650    
 600          10/27/17          1.000996    
 600          10/30/17           .995000    
 600          10/31/17           .996176

查询结果

 Account      Date             ChangePct       Value
 547          10/24/17           .997715     .997715    (.997715 * 1)
 547          10/26/17           .973493     .971268    (.973493 * .997715)
 547          10/27/17          1.001300     .972531    (1.001300 * .971268)
 547          10/30/17           .999541     .972084    (.999541 * .972531)
 600          10/26/17          1.004650    1.004650    (1.004650 * 1)
 600          10/27/17          1.000996    1.005651    (1.000996 * 1.004650)
 600          10/30/17           .995000    1.000622    (.995000 * 1.005651)
 600          10/31/17           .996176     .996796    (.996176  * 1.000622)

Answer 1

<强> SCHEMA：

CREATE TABLE test (
  account INT,
  date1 date,
  change_pct decimal(7,6)
);

INSERT INTO test (account,date1,change_pct) VALUES (547, STR_TO_DATE("10/24/17", "%m/%d/%Y"), .997715); 
INSERT INTO test (account,date1,change_pct) VALUES (547, STR_TO_DATE("10/26/17", "%m/%d/%Y"), .973493);     
INSERT INTO test (account,date1,change_pct) VALUES (547, STR_TO_DATE("10/27/17", "%m/%d/%Y"), 1.001300);     
INSERT INTO test (account,date1,change_pct) VALUES (547, STR_TO_DATE("10/30/17", "%m/%d/%Y"), .999541);     
INSERT INTO test (account,date1,change_pct) VALUES (600, STR_TO_DATE("10/26/17", "%m/%d/%Y"), 1.004650);    
INSERT INTO test (account,date1,change_pct) VALUES (600, STR_TO_DATE("10/27/17", "%m/%d/%Y"), 1.000996 );   
INSERT INTO test (account,date1,change_pct) VALUES (600, STR_TO_DATE("10/30/17", "%m/%d/%Y"), .995000);    
INSERT INTO test (account,date1,change_pct) VALUES (600, STR_TO_DATE("10/31/17", "%m/%d/%Y"), .996176);

<强> QUERY：

select 
    account,
    date1,
    change_pct,
    @prev_acct as ss,
    case 
        when @prev_acct = 1 then 
            @prev_acct:=account 
        when @prev_acct = account then
            @prev_acct:=account
        when @prev_acct != account then
            @prev_acct:=account and @multi:=1
        else 
            @prev_acct:=account
    end as s,
    (@multi := change_pct * @multi) as value
from
    test,
    (select @multi:=1, @prev_acct:= 1) t1
order by
    account,
    date1;

DB FIDDLE

SQL Server乘以按帐户分组的上一行值

1 个答案: