Duplicate entries with unmatching related data
-
Hi,
I am looking for some help on a sql for a report that I need to run. Here is a sample of the data that I need to format
Table One
ID Address
12 500 somewhare NY
13 500 somewhare NY
14 500 somewhare NY
15 505 somewhare NJ
16 505 somewhare NJ
17 506 newtown NH
18 506 newtown NH
Table Two
ID Agent Name
12 Jess
13 Null
14 Jess
15 Dave
16 Dave
17 John
18 Null
Table Three
ID MLSNO
12 Null
13 12313
14 12313
15 12333
16 12333
17 11111
18 Null
I need to get unique address that correspond with the most data. The following sql works however it takes 56 seconds to run and that is too long for the application (especially when the end result is only about 1500 records) I have been racking my head on this for a while. Any help is appreciated.
SET NOCOUNT ON
CREATE TABLE #TMP (lDOCID INT, ADDRESS VARCHAR(40), MLSNO VARCHAR(40), AGENTID VARCHAR(40), FULLNAME VARCHAR(40)) DECLARE @ad VARCHAR(40) DECLARE CUR CURSOR FOR SELECT DISTINCT DSUSER.tblCFD_ADDRESS.sVALUE FROM DSUSER.tblCFD_ADDRESS ORDER BY DSUSER.tblCFD_ADDRESS.sVALUE OPEN CUR FETCH NEXT FROM CUR INTO @ad WHILE @@FETCH_STATUS = 0 BEGIN
INSERT #TMP
SELECT TOP 1 A.lDOCID, A.sVALUE AS ADDRESS, M.sVALUE AS MLSNO,
AG.sVALUE AS AGENTID, USR.sFULLNAME AS FULLNAME FROM DSUSER.tblCFD_ADDRESS A
LEFT JOIN DSUSER.tblCFD_MLS_NO M ON A.lDOCID = M.lDOCID
LEFT JOIN DSUSER.tblCFD_AGENTID AG ON A.lDOCID = AG.lDOCID
LEFT JOIN DSUSER.tblUSER USR ON AG.sVALUE=USR.sNAME
WHERE A.sVALUE = @ad
ORDER BY A.sVALUE, M.sVALUE DESC, AG.sVALUE DESC
FETCH NEXT FROM CUR
INTO @ad
END
CLOSE CUR
DEALLOCATE CUR
SET NOCOUNT OFF
SELECT * FROM #TMP
DROP TABLE #TMP
-
I just relized I posted the actual sql this one is modified to match the sample data posted:
SET NOCOUNT ON
CREATE TABLE #TMP (lDOCID INT, ADDRESS VARCHAR(40), MLSNO VARCHAR(40), AGENTID VARCHAR(40)) DECLARE @ad VARCHAR(40)
DECLARE CUR CURSOR FOR SELECT DISTINCT ADDRESS FROM
ORDER BY ADDRESS
OPEN CUR
FETCH NEXT FROM CUR INTO @ad WHILE @@FETCH_STATUS = 0 BEGIN
INSERT #TMP
SELECT TOP 1 ID, ADDRESS, MLSNO, AGENTID FROM
A
LEFT JOIN
AG ON A.ID = AG.ID
LEFT JOIN
M ON A.ID = M.ID
WHERE A.ADDRESS = @ad
ORDER BY A.sVALUE, M.sVALUE DESC, AG.sVALUE DESC
FETCH NEXT FROM CUR
INTO @ad
END
CLOSE CUR
DEALLOCATE CUR
SET NOCOUNT OFF
SELECT * FROM #TMP
DROP TABLE #TMP
-
Have you tried this?
CREATE
TABLE #TMP
(
RecID INT IDENTITY(1, 1),
lDOCID INT,
ADDRESS VARCHAR(40),
MLSNO VARCHAR(40),
AGENTID VARCHAR(40)
)
INSERT #TMP
SELECT ID,
ADDRESS,
MLSNO,
AGENTID
FROM
A
LEFT JOIN
AG ON A.ID = AG.ID
LEFT JOIN
M ON A.ID = M.ID
ORDER BY A.address,
M.mlsno DESC,
AG.agentid DESC
Select t.lDOCID,
t.ADDRESS,
t.MLSNO,
t.AGENTID
from #tmp AS t
INNER JOIN (
SELECT min(recid) as firstid
from #temp
group by address
) AS d ON d.FirstID = t.RecID
order by t.recid
N 56°04'39.16"
E 12°55'05.25" -
Well, the reason for your query taking so long is because you've written a cursor to handle a task where it is clearly not needed. This can be sped up significantly if it is re-written using SET based logic.
I would normally have done this for you, but it seems that your test code does not work with your test data as you've omitted the sValue column from the test tables. Can you update your post to include the proper sValues for your data? This should be an easy re-write. I'm heading out for the day, but I'll check in tomorrow and if someone else hasn't alredy taken care of you, I will.
-
I know that there is another way to do this not using a cursor but I have a complete block on how it can be done. I will try the post above and see if it works out. If you have another way let me know. I have replaced the sValues.
SET NOCOUNT ON
CREATE TABLE #TMP (lDOCID INT, ADDRESS VARCHAR(40), MLSNO VARCHAR(40), AGENTID VARCHAR(40)) DECLARE @ad VARCHAR(40)
DECLARE CUR CURSOR FOR SELECT DISTINCT ADDRESS FROM
ORDER BY ADDRESS
OPEN CUR
FETCH NEXT FROM CUR INTO @ad WHILE @@FETCH_STATUS = 0 BEGIN
INSERT #TMP
SELECT TOP 1 ID, ADDRESS, MLSNO, AGENTID FROM
A
LEFT JOIN
AG ON A.ID = AG.ID
LEFT JOIN
M ON A.ID = M.ID
WHERE A.ADDRESS = @ad
ORDER BY Address, MLSNO DESC, AGENTIT DESC
FETCH NEXT FROM CUR
INTO @ad
END
CLOSE CUR
DEALLOCATE CUR
SET NOCOUNT OFF
SELECT * FROM #TMP
DROP TABLE #TMP
-
The message from peter worked great. Thank You.
I am interested in how you would use set based logic to do this. Can you explain by either using the sample I provided or something simular
-
I am using set based logic.
Your original code took 56 seconds to complete for 1500 records. If you try my approach, how many seconds does that take?
N 56°04'39.16"
E 12°55'05.25" -
.6 seconds
Ok so then I have used set based logic in the past just never knew that was what it was called. Thank you.
-
-- prepare sample data
declare
@tableone table (id int, address varchar(40))
insert
@tableone
select
12, '500 somewhare NY' union all
select
13, '500 somewhare NY' union all
select
14, '500 somewhare NY' union all
select
15, '505 somewhare NJ' union all
select
16, '505 somewhare NJ' union all
select
17, '506 newtown NH' union all
select
18, '506 newtown NH'
declare
@tabletwo table (id int, agentname varchar(40))
insert
@tabletwo
select
12, 'Jess' union all
select
13, null union all
select
14, 'Jess' union all
select
15, 'Dave' union all
select
16, 'Dave' union all
select
17, 'John' union all
select
18, null
declare
@tablethree table (id int, mlsno varchar(40))
insert
@tablethree
select
12, null union all
select
13, '12313' union all
select
14, '12313' union all
select
15, '12333' union all
select
16, '12333' union all
select
17, '11111' union all
select
18, null
-- stage the data
declare
@stage table (recid int identity, ldocid int, address varchar(40), mlsno varchar(40), agentname varchar(40))
insert
@stage (ldocid, address, mlsno, agentname)
select
a.id,
a
.address,
m
.mlsno,
ag
.agentname
from
@tableone as a
left
join @tabletwo as ag on ag.id = a.id
left
join @tablethree as m on m.id = a.id
order
by a.address,
m
.mlsno desc,
ag
.agentname desc
-- show the expected output
select
s.ldocid,
s
.address,
s
.mlsno,
s
.agentname
from
@stage as s
inner
join (
select min(recid) as firstid
from @stage
group by address
) as d on d.firstid = s.recid
order
by s.recid
ldocid address mlsno agentname
14 500 somewhare NY 12313 Jess
15 505 somewhare NJ 12333 Dave
17 506 newtown NH 11111 John
N 56°04'39.16"
E 12°55'05.25" -
Going with <<I need to get unique address that correspond with the most data>>, this will preferentially pick an ID/address line which joins to non-null MLSNO and agent...
DROP TABLE #TableOne CREATE TABLE #TableOne (ID int, Address VARCHAR(20))
INSERT INTO #TableOne (ID, Address) SELECT 12, '500 somewhare NY' UNION ALL SELECT 13, '500 somewhare NY' UNION ALL SELECT 14, '500 somewhare NY' UNION ALL SELECT 15, '505 somewhare NJ' UNION ALL SELECT 16, '505 somewhare NJ' UNION ALL SELECT 17, '506 newtown NH' UNION ALL SELECT 18, '506 newtown NH'
--SELECT * FROM #TableOne
DROP TABLE #TableTwo CREATE TABLE #TableTwo (ID int, AgentName VARCHAR(4))
INSERT INTO #TableTwo (ID, AgentName) SELECT 12, 'Jess' UNION ALL SELECT 13, Null UNION ALL SELECT 14, 'Jess' UNION ALL SELECT 15, 'Dave' UNION ALL SELECT 16, 'Dave' UNION ALL SELECT 17, 'John' UNION ALL SELECT 18, Null
--SELECT * FROM #TableTwo
DROP TABLE #TableThree CREATE TABLE #TableThree (ID int, MLSNO int)
INSERT INTO #TableThree (ID, MLSNO) SELECT 12, Null UNION ALL SELECT 13, 12313 UNION ALL SELECT 14, 12313 UNION ALL SELECT 15, 12333 UNION ALL SELECT 16, 12333 UNION ALL SELECT 17, 11111 UNION ALL SELECT 18, Null
--SELECT * FROM #TableThree
DROP TABLE #CalcTable SELECT ad.ID, ad.Address, mls.MLSNO, ag.AgentName, CASE WHEN mls.MLSNO IS NULL THEN 0 ELSE 1 END + CASE WHEN ag.AgentName IS NULL THEN 0 ELSE 1 END AS ValueCount INTO #CalcTable FROM #TableOne ad INNER JOIN #TableTwo ag ON ag.ID = ad.ID INNER JOIN #TableThree mls ON mls.ID = ag.ID
--SELECT * FROM #CalcTable
SELECT ad.ID, ad.Address, mls.MLSNO, ag.AgentName FROM #TableOne ad LEFT JOIN #TableTwo ag ON ag.ID = ad.ID LEFT JOIN #TableThree mls ON mls.ID = ag.ID INNER JOIN ( SELECT MAX(ID) AS ID FROM #CalcTable a INNER JOIN (SELECT Address, MAX(ValueCount) AS MAXValueCount FROM #CalcTable
GROUP BY Address) b ON b.Address = a.Address AND b.MAXValueCount = a.ValueCount GROUP BY a.Address ) t ON t.ID = ad.ID ORDER BY ad.ID, mls.MLSNO DESC, ag.AgentName DESC
Cheers
ChrisM
“Write the query the simplest way. If through testing it becomes clear that the performance is inadequate, consider alternative query forms.” - Gail ShawFor fast, accurate and documented assistance in answering your questions, please read this article.
Understanding and using APPLY, (I) and (II) Paul White
Hidden RBAR: Triangular Joins / The "Numbers" or "Tally" Table: What it is and how it replaces a loop Jeff Moden
Viewing 10 posts - 1 through 9 (of 9 total)
You must be logged in to reply to this topic. Login to reply