Return rows that have a negative equivalent - sql-server

I have a non-normalized table with several columns. I would like to return all columns that have a positive number along with a negative number of the same value.
Example:
ID | Value
-------------
1 | 10
1 | -10
3 | 15
3 | 15
4 | -1
5 | 4
Current Output:
ID | Values
-------------
1 | 10
1 | -10
3 | 15
3 | 15
Desired Output:
ID | Value
-------------
1 | 10
1 | -10
I have made a windows function as seen below that will select absolute values that are the same, but this includes pairs where there are a positive number.
select Count(*) Over (Partition By DVN, [Tran Date], [Reference Number],Description,Vendor, Abs([Maintenance Expense])) As cnt , *
From WorkTemp.dbo.Customer2700Combine
Where [Maintenance Expense] Is Not Null
Order By 1 Desc,DVN, [Tran Date], [Reference Number],Description,Vendor, Abs([NonRental Total])

Not sure if your requirement is by [ID], looking at your example, description and desired output, this is how I would do it:
DROP TABLE IF EXISTS #sopg;
SELECT [ID],
[VALUE]
INTO #sopg
FROM
(
SELECT 1 AS ID,
10 AS VALUE
UNION
SELECT 1 AS ID,
-10 AS VALUE
UNION
SELECT 3 AS ID,
15 AS VALUE
UNION
SELECT 3 AS ID,
15 AS VALUE
UNION
SELECT 4 AS ID,
-1 AS VALUE
UNION
SELECT 5 AS ID,
4 AS VALUE
) x;
-- Assuming that one ID can only have maximum 2 rows (like your example above) and want this by ID
SELECT s.[ID],
s.[VALUE]
FROM #sopg s
INNER JOIN
(
SELECT ID,
SUM(VALUE) SumZero
FROM #sopg
GROUP BY ID
HAVING SUM(VALUE) = 0
) SumZero ON SumZero.ID = s.ID
-- Another way, assuming that ID can have more than 2 rows and different values
DROP TABLE IF EXISTS #sopg2;
SELECT [ID],
[VALUE]
INTO #sopg2
FROM
(
SELECT 1 AS ID,
10 AS VALUE
UNION
SELECT 1 AS ID,
-10 AS VALUE
UNION
SELECT 1 AS ID,
-9 AS VALUE
UNION
SELECT 3 AS ID,
15 AS VALUE
UNION
SELECT 3 AS ID,
15 AS VALUE
UNION
SELECT 4 AS ID,
-1 AS VALUE
UNION
SELECT 5 AS ID,
4 AS VALUE
) x
SELECT a.[ID],
a.[VALUE]
FROM #sopg2 a
INNER JOIN #sopg b ON b.ID = a.ID AND a.VALUE = -b.VALUE

Related

Number rows based on id

I have a query that returns something like this:
Id | Value
1 | Hi,
1 | I'm
2 | just
2 | an
2 | example
3 | message.
What I want to do is number the rows based on the id. So with the example above, I want to return something like this:
Id | Value | Number
1 | Hi, | 1
1 | I'm | 2
2 | just | 1
2 | an | 2
2 | example | 3
3 | message. | 1
Is there a simple way to do this in the same query as the one you would use in the first example?
The problem is there is NOTHING in that table that guarantees the order of rows, so the result you want cannot be guaranteed.
select *
, row_number() over(partition by id order by (select 1)) as Number
from yourtable
You can make use of the rownumber function.
SELECT ROW_NUMBER OVER (PARTITION BY ID ORDER BY (SELECT 1)) AS RowNumb
You Can Use the ROW_NUMBER() Function and Partition by Id. Like this
DECLARE #MyTable AS TABLE
(
Id INT,
[Value] VARCHAR(50)
)
INSERT INTO #MyTable
(
Id,
[Value]
)
SELECT '1','Hi,' UNION
SELECT '1','I''m' UNION
SELECT '2','just' UNION
SELECT '2','an' UNION
SELECT '2','example' UNION
SELECT '3','message.'
SELECT
*,
Number = ROW_NUMBER() OVER(PARTITION BY Id ORDER BY Id)
FROM #MyTable
Result
Id Value Number
----------- -------------------------------------------------- --------------------
1 Hi, 1
1 I'm 2
2 an 1
2 example 2
2 just 3
3 message. 1
Please try following script.
DECLARE #MyTable AS TABLE
(
Id INT,
[Value] VARCHAR(50)
)
INSERT INTO #MyTable
(
Id,
[Value]
)
SELECT '1','Hi,' UNION
SELECT '1','I''m' UNION
SELECT '2','just' UNION
SELECT '2','an' UNION
SELECT '2','example' UNION
SELECT '3','message.'
select *, dense_rank()over(partition by Id order by Value) as Number from #MyTable
/*
Id Value Number
----------- -------------------------------------------------- --------------------
1 Hi, 1
1 I'm 2
2 an 1
2 example 2
2 just 3
3 message. 1
*/
Best Regards,
Rachel

Grouping in SQL Server

I have data similar to below. I need to group the status column based on the occurrence of data.
Id Status Value
1 K 1
2 K 3
3 K 2
4 B 2
5 B 3
6 K 6
7 J 5
8 J 2
I want data as below
Status Value
K 6
B 5
K 6
J 7
I need the cumulative sum of the value column.In the status column if the data is same consecutively, then I need to add the value columns. I cannot apply group by. In the example given K is repeated twice, because they are not consecutive.
I have tried below query, but it doesn't work as required.
select Status,
(select sum(value)
from table t2
where
t2.Status = t.Status and
t2.SNO <= t.SNO
) as total
from table t;
This is a Gaps and Islands Question
I tackle these by using the incrementing Id and combing this with ROW_NUMBER window function
--Using a CTE just to replicate the sample data
;WITH cteX (Id,Status,Value)
AS(
SELECT 1,'K', 1 UNION ALL
SELECT 2,'K', 3 UNION ALL
SELECT 3,'K', 2 UNION ALL
SELECT 4,'B', 2 UNION ALL
SELECT 5,'B', 3 UNION ALL
SELECT 6,'K', 6 UNION ALL
SELECT 7,'J', 5 UNION ALL
SELECT 8,'J', 2
)
SELECT
Grp = Id - ROW_NUMBER()OVER( PARTITION BY X.Status ORDER BY X.Id)
, X.Id
, X.Status
, X.Value
FROM
cteX X
ORDER BY
X.Id
This gives this result set, note the Grp column
Grp Id Status Value
------- ------- ------- -------
0 1 K 1
0 2 K 3
0 3 K 2
3 4 B 2
3 5 B 3
2 6 K 6
6 7 J 5
6 8 J 2
Then combine with a CTE or derived table you can get your expected output
--Using a CTE just to replicate the sample data
;WITH cteX (Id,Status,Value)
AS(
SELECT 1,'K', 1 UNION ALL
SELECT 2,'K', 3 UNION ALL
SELECT 3,'K', 2 UNION ALL
SELECT 4,'B', 2 UNION ALL
SELECT 5,'B', 3 UNION ALL
SELECT 6,'K', 6 UNION ALL
SELECT 7,'J', 5 UNION ALL
SELECT 8,'J', 2
)
SELECT Y.Status
, Value = SUM(Y.Value)
FROM
(
SELECT TOP 100 PERCENT
Grp = Id - ROW_NUMBER()OVER( PARTITION BY X.Status ORDER BY X.Id)
, X.Id
, X.Status
, X.Value
FROM
cteX X
ORDER BY
X.Id
) Y
GROUP BY
Y.Grp, Y.Status
Output
Status Value
------- -------
B 5
J 7
K 6
K 6
Update Question include "Preserve order" solution
Just include an Order by MIN(Id)
--Using a CTE just to replicate the sample data
;WITH cteX (Id,Status,Value)
AS(
SELECT 1,'K', 1 UNION ALL
SELECT 2,'K', 3 UNION ALL
SELECT 3,'K', 2 UNION ALL
SELECT 4,'B', 2 UNION ALL
SELECT 5,'B', 3 UNION ALL
SELECT 6,'K', 6 UNION ALL
SELECT 7,'J', 5 UNION ALL
SELECT 8,'J', 2
)
SELECT
Y.[Status]
,[Value] = SUM(Y.[Value])
FROM
(
SELECT
Grp = Id - ROW_NUMBER()OVER( PARTITION BY X.[Status] ORDER BY X.Id)
, X.Id
, X.[Status]
, X.[Value]
FROM
cteX X
) Y
GROUP BY
Y.Grp, Y.[Status]
ORDER BY
MIN(Y.Id) --preserve the Status Order
Output
Status Value
------- -------
K 6
B 5
K 6
J 7

how to select rows where column value has changed

I have a table in which I have few columns like below:
Cusnbr Name LoadNumber
1 Z 10
1 Z 9
1 Z 8
1 C 7
1 C 6
1 C 5
1 B 4
1 B 3
1 A 2
1 A 1
it is just for one cusnbr there are million of cusnbr like this..
I want output like below
Cusnbr Name LoadNumber
1 C 7
1 B 4
1 A 2
For that I write below query in sql server 2008:
;With x as
(
Select * ,rn=Row_number() over (order by cusnbr,loadnumber) from table
)
select x.* from x left outer join x as y on x.rn=y.rn+1
and x.name<>y.name where y.name is not null
but I am not getting the desired output in the above code I am getting last Z also which I don't want and I am getting irregular data not in the correct form in which I want
Any help will be appreciated !!
like this I want but not able to get the desired output
I use this example
Though the question is not clear to me , Guessing from the output I have tried out Dense Rank . I guessed you want the record with highest LoadNumber with the same name .
Select * from cteTrial where LoadNumber in (
Select MAX(x.LoadNumber) as LoadNumber from (
Select cusnbr , name , LoadNumber , DENSE_RANK() over (order by Name desc )
as Dense from cteTrial) as x group by x.Dense
)
If you can use CTE it will produce better performance .
i written the code as per expected
;With cte(Cusnbr , Name , LoadNumber)
AS
(
SELECT 1,'Z', 10 Union all
SELECT 1,'Z', 9 Union all
SELECT 1,'Z', 8 Union all
SELECT 1,'C', 7 Union all
SELECT 1,'C', 6 Union all
SELECT 1,'C', 5 Union all
SELECT 1,'B', 4 Union all
SELECT 1,'B', 3 Union all
SELECT 1,'A', 2 Union all
SELECT 1,'A', 1
)
SELECT cusnbr,
NAME,
loadnumber
FROM (SELECT *,
Row_number()
OVER(
partition BY NAME
ORDER BY loadnumber DESC) AS RNk,
Row_number()
OVER(
ORDER BY (SELECT 1)) - 1 AS RNO
FROM (SELECT *
FROM cte)dt)DT2
WHERE DT2.rnk = 1
AND rno > 0
ORDER BY NAME DESC
Result
cusnbr NAME loadnumber
-------------------------
1 C 7
1 B 4
1 A 2

Make query like treeview

Brothers can you help me? Thanks
Table A
Id Name IdParent
1 Operation Null
2 Developer 1
3 Android 2
4 IOS 2
Expectes result:
ID Name
1 +Operation
2 +------ Developer
3 +------------Android
4 +------------ IOS
By adding a sequence during the recursive build, you can easily create the proper presentation sequence and nesting
Declare #YourTable table (id int,IdParent int,Name varchar(50))
Insert into #YourTable values
( 1, NULL,'Operation')
,( 2, 1 ,'Developer')
,( 3, 2 ,'Android')
,( 4, 2 ,'IOS')
,( 5, 1 ,'Poet')
,( 6, 5 ,'Limerick')
,( 7, 5 ,'Haiku')
Declare #Top int = null --<< Sets top of Hier Try 2
Declare #Nest varchar(25) = '|-----' --<< Optional: Added for readability
;with cteP as (
Select Seq = cast(10000+Row_Number() over (Order by Name) as varchar(500))
,ID
,IdParent
,Lvl=1
,Name
From #YourTable
Where IsNull(#Top,-1) = case when #Top is null then isnull(IdParent ,-1) else ID end
Union All
Select Seq = cast(concat(p.Seq,'.',10000+Row_Number() over (Order by r.Name)) as varchar(500))
,r.ID
,r.IdParent
,p.Lvl+1
,r.Name
From #YourTable r
Join cteP p on r.IdParent = p.ID)
Select A.ID
,A.IdParent
,A.Lvl
,Name = Replicate(#Nest,A.Lvl-1) + A.Name
From ctep A
Order By A.Seq
Returns
ID IdParent Lvl Name
1 NULL 1 Operation
2 1 2 |-----Developer
3 2 3 |-----|-----Android
4 2 3 |-----|-----IOS
5 1 2 |-----Poet
7 5 3 |-----|-----Haiku
6 5 3 |-----|-----Limerick
Here's another version:
WITH RawData AS (
SELECT 1 AS Id, 'Operation' AS Name, CONVERT(INT, NULL) AS IdParent
UNION ALL
SELECT 2 AS Id, 'Developer' AS Name, 1 AS IdParent
UNION ALL
SELECT 3 AS Id, 'Android' AS Name, 2 AS IdParent
UNION ALL
SELECT 4 AS Id, 'IOS' AS Name, 2 AS IdParent),
Depth AS (
SELECT
Id,
1 AS depth,
IdParent
FROM
RawData
UNION ALL
SELECT
d.Id,
d.depth + 1,
r.IdParent
FROM
Depth d
INNER JOIN RawData r ON r.Id = d.IdParent),
MaxDepth AS (
SELECT
Id,
MAX(depth) AS depth
FROM
Depth
GROUP BY
Id)
SELECT
r.Id,
'+' + REPLICATE('----', m.depth - 1) + r.Name AS Name
FROM
RawData r
INNER JOIN MaxDepth m ON m.Id = r.Id;
Results:
Id Name
1 +Operation
2 +----Developer
3 +--------Android
4 +--------IOS
DECLARE #mockup TABLE(Id INT, Name VARCHAR(100), IdParent INT);
INSERT INTO #mockup VALUES
(1,'Operation',Null)
,(2,'Developer',1)
,(3,'Android',2)
,(4,'IOS',2);
--The query uses a recursive CTE and finally REPLICATE with the recursive level to add the number of hyphens...
WITH recCTE AS
(
SELECT Id, Name, 1 AS Lvl, CAST(REPLACE(STR(ROW_NUMBER() OVER (ORDER BY Id),5),' ','0') AS VARCHAR(MAX)) AS Seq
FROM #mockup
WHERE IdParent IS NULL
UNION ALL
SELECT m.Id,m.Name,r.Lvl +1,r.Seq + '.' + REPLACE(STR(ROW_NUMBER() OVER (ORDER BY m.Id),5),' ','0')
FROM #mockup AS m
INNER JOIN recCTE AS r ON m.IdParent=r.Id
)
SELECT *
,'+' + REPLICATE('-',Lvl*4) + Name
FROM recCTE
ORDER BY Seq
the result
+----+-----------+-----+----------------------+
| Id | Name | Lvl | (Kein Spaltenname) |
+----+-----------+-----+----------------------+
| 1 | Operation | 1 | +----Operation |
+----+-----------+-----+----------------------+
| 2 | Developer | 2 | +--------Developer |
+----+-----------+-----+----------------------+
| 3 | Android | 3 | +------------Android |
+----+-----------+-----+----------------------+
| 4 | IOS | 3 | +------------IOS |
+----+-----------+-----+----------------------+

Right using of PIVOT-statement in SQL Server

I have an audit table, part of data is looked like:
Id ColumnName Value RowId
---------------------------------
1 EditCheckId 100 1
2 FieldData_Id 10 1
3 EditType 1 1
4 Outcome True 1
5 EditCheckId 200 2
6 FieldData_Id 20 2
7 EditType 2 2
8 Outcome False 2
9 EditCheckId 300 3
10 FieldData_Id 30 3
11 EditType 3 3
12 Outcome True 3
and I want to construct such table groupping data by RowId
EditCheck_Id FieldData_Id EditType Outcome
---------------------------------------------------
100 10 1 True
200 20 2 False
300 30 3 True
I've tried the query:
select [EditCHeck_Id], [FieldData_Id], [EditType], [Outcome]
from
(
select [ColumnName], [Value]
from Audit a
) as [SourceTable]
pivot
(
max([Value])
for [ColumnName] in ([EditCHeck_Id], [FieldData_Id], [EditType], [Outcome])
) as [PivotTable];
http://sqlfiddle.com/#!6/7af71/3
using PIVOT statement but there is only one row in answer. Where is my problem?
You'll need some value to GROUP BY to make each of the rows distinct. Typically, you would use a windowing function like row_number() to generate a unique sequenced number over your current columnname. You can alter your query to the following:
select
[EditCheckId],
[FieldData_Id],
[EditType],
[Outcome]
from
(
select [ColumnName], [Value],
rn = row_number() over(partition by ColumnName order by id)
from Audit a
) as [SourceTable]
pivot
(
max([Value])
for [ColumnName] in ([EditCheckId], [FieldData_Id], [EditType], [Outcome])
) as [PivotTable];
See SQL Fiddle with Demo.
You could also use an aggregate function with a CASE expression to get the final result:
select
max(case when ColumnName = 'EditCheckId' then value end) [EditCheckId],
max(case when ColumnName = 'FieldData_Id' then value end) [FieldData_Id],
max(case when ColumnName = 'EditType' then value end) [EditType],
max(case when ColumnName = 'Outcome' then value end) [Outcome]
from
(
select [ColumnName], [Value],
rn = row_number() over(partition by ColumnName order by id)
from Audit a
) d
group by rn;
See SQL Fiddle with Demo Both give the result:
| EDITCHECKID | FIELDDATA_ID | EDITTYPE | OUTCOME |
|-------------|--------------|----------|---------|
| 100 | 10 | 1 | True |
| 200 | 20 | 2 | False |
| 300 | 30 | 3 | True |
You need something to distinguish the output rows, like for example your RowId
pivot does the grouping for you like so:
select [RowID], [EditCheckId], [FieldData_Id], [EditType], [Outcome]
from
(
select [ColumnName], [Value], [RowId]
from Audit a
) as [SourceTable]
pivot
(
max([Value])
for [ColumnName] in ([EditCheckId], [FieldData_Id], [EditType], [Outcome])
) as [PivotTable];
edited Fiddle demo

Resources