Update the sum value in second row - sql

Question

I have table like followed by
User Col1 Col2
ABC 35 75
ABC 500 75
and I need the select query for following output.
User Col1 Col2 Result
ABC 35 75 40
ABC 500 75 115
Check col1<col2 then Result Col2-Col1
else
Col2 + Result(Row1)
Ex:
35<75 then 40
else
75+40
I don't know how to start please help us for output..
Thanks in advance
What I have tried
Select User,Col1,Col2
,Case When Col1<Col2 then Col2-Col1 Else Col2 End as Result
Then i got the output like followed by
User Col1 Col2 Result
ABC 35 75 40
ABC 500 75 **75**
I need 115 instead of 75 in Result of second row

Answer 1

CREATE TABLE #tmp
(
UserName VARCHAR(10),
col1 INT,
col2 INT
)
INSERT INTO #tmp (UserName, col1, col2)
SELECT 'ABC', 35, 75
UNION ALL SELECT 'ABC', 500, 75
SELECT tmp.UserName
,tmp.col1
,tmp.col2
,CASE WHEN tmp.Result = 0 THEN tmp.col2 + LAG(tmp.Result) OVER (ORDER BY (SELECT NULL)) ELSE tmp.Result END AS Result
FROM
(
Select UserName,Col1,Col2
,Case When Col1<Col2 then Col2-Col1 Else 0 End as Result
FROM #tmp
)tmp

Answer 2

You could try something like this:
SELECT *
,Case When Col1 < Col2 then Col2-Col1
Else (Col2 + LAG(Col2-Col1 ,1) OVER(ORDER BY Col2)) --Order by
End as Result
FROM TableName
By using LAG you can access the previous row of data. Then all I do is add the previous result to the Col2 value.
Else (Col2 + LAG(Col2-Col1 ,1) OVER(ORDER BY Col2))
Important note below.
The only thing to note here is the ORDER BY Col2. At the moment that will work for just the two rows of data. But, if you have more you should add some sort of auto incrementing ID field to keep the order the same as how you enter the data. Then you would change it to ORDER BY ID.

Answer 3

You can try below using lag() function
Select User,Col1,Col2
,Case When Col1<Col2 then Col2-Col1
Else Col2+lag(col2-col1) over(partition by user order by col2 ) End as Result
from tablename

Answer 4

You need to use the LAG function for this.here is the code...if you like the answer please vote.
Select *
,Case When Col1<Col2 then Col2-Col1
Else Col2+lag(col2-col1) over(partition by user order by col2 ) End as Result
from review

Answer 5

First, you need a column that represents the ordering of the rows.
If I generalize the problem, then you need the groups defined by col1 < col2. You can define the groups by taking a cumulative sum of when col1 < col2.
The actual calculation is then a cumulative sum over the group:
select t.*,
(case when col1 < col2 then col2 - col1
else sum(case when col1 > col2 then col2 - col1 else col2 end) over (partition by user, grp order by ?)
end) as result
from (select t.*,
sum(case when col1 < col2 then 1 else 0 end) over (partition by user order by ?) as grp
from t
) t
from t;
The ? is for the column that specifies the ordering of the rows.