find a duplicate series in SQL - sql-server-2008

Question

I have a table with 3 columns containing a variable number of records based off of the first column which is a foreign key. I am trying to determine if I can detect when there is a duplicate across multiple rows for an entire series
declare #finddupseries table
(
portid int,
asset_id int,
allocation float
)
;
INSERT INTO #finddupseries
SELECT 250, 6, 0.05 UNION ALL
SELECT 250, 66, 0.8 UNION ALL
SELECT 250, 2, 0.105 UNION ALL
SELECT 250, 4, 0.0225 UNION ALL
SELECT 250, 5, 0.0225 UNION ALL
SELECT 251, 13, 0.6 UNION ALL
SELECT 251, 2, 0.3 UNION ALL
SELECT 251, 5, 0.1 UNION ALL
SELECT 252, 13, 0.8 UNION ALL
SELECT 252, 2, 0.15 UNION ALL
SELECT 252, 5, 0.05 UNION ALL
SELECT 253, 13, 0.4 UNION ALL
SELECT 253, 2, 0.45 UNION ALL
SELECT 253, 5, 0.15 UNION ALL
SELECT 254, 6, 0.05 UNION ALL
SELECT 254, 66, 0.8 UNION ALL
SELECT 254, 2, 0.105 UNION ALL
SELECT 254, 4, 0.0225 UNION ALL
SELECT 254, 5, 0.0225
select * from #finddupseries
The records for portid 250 and 254 match.
Is there any way I can write a query to detect this?
edit: yes, the entire series must match. Also, if there was a way to determine which one it DID match would be helpful as the actual table has around 10k records.
thanks!

Answer 1

This query will give you all the values converted into a string grouped by port_id
SELECT fus1.portid,
(
SELECT CONVERT (VARCHAR, fus2.asset_id) + CONVERT (VARCHAR, fus2.allocation) + ','
FROM #finddupseries fus2
WHERE 1=1
AND fus1.portid = fus2.portid
ORDER BY fus2.portid, fus2.asset_id, fus2.allocation
FOR XML PATH ('')
) AllValuesFromAllRows
FROM #finddupseries fus1
GROUP BY fus1.portid
the output should look like this
portid AllValuesFromAllRows
----------- ------------------------------------------------------
250 20.105,40.0225,50.0225,60.05,660.8,
251 20.3,50.1,130.6,
252 20.15,50.05,130.8,
253 20.45,50.15,130.4,
254 20.105,40.0225,50.0225,60.05,660.8,
Now, lets do a group by with a having!
;With DuplicateFinder as
(
SELECT fus1.portid,
(
SELECT CONVERT (VARCHAR, fus2.asset_id) + CONVERT (VARCHAR, fus2.allocation) + ','
FROM #finddupseries fus2
WHERE 1=1
AND fus1.portid = fus2.portid
ORDER BY fus2.portid, fus2.asset_id, fus2.allocation
FOR XML PATH ('')
) AllValuesFromAllRows
FROM #finddupseries fus1
GROUP BY fus1.portid
)
SELECT AllValuesFromAllRows, COUNT (*) NumDups
FROM DuplicateFinder
GROUP BY AllValuesFromAllRows
Having COUNT (*) > 1
You should get
AllValuesFromAllRows NumDups
----------------------------------------------- -----------
20.105,40.0225,50.0225,60.05,660.8, 2
So here is everything put together
SET NOCOUNT ON
declare #finddupseries table
(
portid int,
asset_id int,
allocation float
)
;
INSERT INTO #finddupseries
SELECT 250, 6, 0.05 UNION ALL
SELECT 250, 66, 0.8 UNION ALL
SELECT 250, 2, 0.105 UNION ALL
SELECT 250, 4, 0.0225 UNION ALL
SELECT 250, 5, 0.0225 UNION ALL
SELECT 251, 13, 0.6 UNION ALL
SELECT 251, 2, 0.3 UNION ALL
SELECT 251, 5, 0.1 UNION ALL
SELECT 252, 13, 0.8 UNION ALL
SELECT 252, 2, 0.15 UNION ALL
SELECT 252, 5, 0.05 UNION ALL
SELECT 253, 13, 0.4 UNION ALL
SELECT 253, 2, 0.45 UNION ALL
SELECT 253, 5, 0.15 UNION ALL
SELECT 254, 6, 0.05 UNION ALL
SELECT 254, 66, 0.8 UNION ALL
SELECT 254, 2, 0.105 UNION ALL
SELECT 254, 4, 0.0225 UNION ALL
SELECT 254, 5, 0.0225
;With PivotAssetIdAndAllocation as
(
SELECT fus1.portid,
(
SELECT CONVERT (VARCHAR, fus2.asset_id) + '_'+ CONVERT (VARCHAR, fus2.allocation) + '~~'
FROM #finddupseries fus2
WHERE 1=1
AND fus1.portid = fus2.portid
ORDER BY fus2.portid, fus2.asset_id, fus2.allocation
FOR XML PATH ('')
) AllValuesFromAllRows
FROM #finddupseries fus1
GROUP BY fus1.portid
)
,
ListOfDuplicates AS
(
SELECT AllValuesFromAllRows, COUNT (*) NumDups
FROM PivotAssetIdAndAllocation
GROUP BY AllValuesFromAllRows
Having COUNT (*) > 1
)
SELECT portid, AllValuesFromAllRows
FROM PivotAssetIdAndAllocation
WHERE AllValuesFromAllRows IN (SELECT AllValuesFromAllRows FROM ListOfDuplicates)
and the output is
portid AllValuesFromAllRows
----------- ----------------------------------------------------------------------
250 2_0.105~~4_0.0225~~5_0.0225~~6_0.05~~66_0.8~~
254 2_0.105~~4_0.0225~~5_0.0225~~6_0.05~~66_0.8~~

Answer 2

this should do the trick:
declare #finddupseries table
(
portid int,
asset_id int,
allocation float
)
;
INSERT INTO #finddupseries
SELECT 250, 6, 0.05 UNION ALL
SELECT 250, 66, 0.8 UNION ALL
SELECT 250, 2, 0.105 UNION ALL
SELECT 250, 4, 0.0225 UNION ALL
SELECT 250, 5, 0.0225 UNION ALL
SELECT 251, 13, 0.6 UNION ALL
SELECT 251, 2, 0.3 UNION ALL
SELECT 251, 5, 0.1 UNION ALL
SELECT 252, 13, 0.8 UNION ALL
SELECT 252, 2, 0.15 UNION ALL
SELECT 252, 5, 0.05 UNION ALL
SELECT 253, 13, 0.4 UNION ALL
SELECT 253, 2, 0.45 UNION ALL
SELECT 253, 5, 0.15 UNION ALL
SELECT 254, 6, 0.05 UNION ALL
SELECT 254, 66, 0.8 UNION ALL
SELECT 254, 2, 0.105 UNION ALL
SELECT 254, 4, 0.0225 UNION ALL
SELECT 254, 5, 0.0225 UNION ALL
SELECT 255, 13, 0.6 UNION ALL
SELECT 255, 2, 0.3 UNION ALL
SELECT 255, 5, 0.1
;with cteGetDupes
as
(
SELECT row_number() over (partition by asset_id, allocation
order by portid desc, asset_id desc, allocation desc)
AS RNDesc
, row_number() over (partition by asset_id, allocation
order by portid, asset_id, allocation)
AS RNAsc
, *
FROM #finddupseries
)
SELECT portid, asset_id, allocation
FROM cteGetDupes
WHERE RNDesc - RNAsc != 0
order by portid, asset_id, allocation