Please start any new threads on our new
site at https://forums.sqlteam.com. We've got lots of great SQL Server
experts to answer whatever question you can come up with.
| Author |
Topic |
|
RyanAustin
Yak Posting Veteran
50 Posts |
Posted - 2009-06-22 : 13:48:23
|
| How would I go about excluding values from a query when I cannot seem to use distinct or group by properly? I have the following:SELECT callid, recvd, emailid, custid, firstname, lastname, loginid, callsource--INTO #CI_SurveyFROM ( SELECT top 50 percent c.callid, getdate() as recvd, p.emailid, p.custid, p.firstname, p.lastname, p.loginid, c.callsource FROM Prod.dbo.calllog c tablesample(1000000 rows) LEFT OUTER JOIN Prod.dbo.[profile] p ON c.custid = p.custid WHERE c.closeddate = CONVERT(VARCHAR(10), GETDATE() -3, 120) AND c.custtype = 'Employee' AND p.loginid <> 'noaddress' AND p.emailid IS NOT NULL AND p.SurveyOptOut is not null ORDER BY checksum(newid()), c.callid) XXXGROUP BY callid, recvd, emailid, custid, firstname, lastname, loginid, callsourceHAVING COUNT(custid) = 1ORDER BY custidThe query pulls a random sample of all trouble tickets closed on a given day (I'm using the -3 because we don't close any on the weekends). I only want distinct custid to show up, and when I run this I get multiple custid. I really don't care about the callid that gets returned, I just don't want to see more then 1 custid per day that I run this. Thanks, Ryan |
|
|
jimf
Master Smack Fu Yak Hacker
2875 Posts |
Posted - 2009-06-22 : 14:31:05
|
| If there are different callids within a custid, then you'll get a record for each callid. If it isn't necessary for your results, just drop it out of your Select statement and Group By statement and you should be okay.Jim |
 |
|
|
webfred
Master Smack Fu Yak Hacker
8781 Posts |
Posted - 2009-06-22 : 15:12:20
|
I can believe that each retrieved record for one custid hase the same emailid, custid, firstname, lastnamebut what is aboutrecvd, callsource, loginid?If you want only one record per custid this would go like this:SELECT max(callid) as callid, max(recvd) as recvd, emailid, custid, firstname, lastname, max(loginid) as loginid, max(callsource) as callsource--INTO #CI_SurveyFROM (SELECT top 50 percent c.callid,getdate() as recvd,p.emailid,p.custid,p.firstname,p.lastname,p.loginid,c.callsourceFROM Prod.dbo.calllog c tablesample(1000000 rows)LEFT OUTER JOIN Prod.dbo.[profile] p ON c.custid = p.custidWHERE c.closeddate = CONVERT(VARCHAR(10), GETDATE() -3, 120)AND c.custtype = 'Employee'AND p.loginid <> 'noaddress'AND p.emailid IS NOT NULLAND p.SurveyOptOut is not nullORDER BY checksum(newid()), c.callid) XXXGROUP BY custid, emailid, firstname, lastnameORDER BY custid No, you're never too old to Yak'n'Roll if you're too young to die. |
 |
|
|
RyanAustin
Yak Posting Veteran
50 Posts |
Posted - 2009-06-22 : 15:58:18
|
| Thanks WebFred, that did the trick. Ryan |
 |
|
|
webfred
Master Smack Fu Yak Hacker
8781 Posts |
Posted - 2009-06-22 : 16:35:57
|
welcome  No, you're never too old to Yak'n'Roll if you're too young to die. |
 |
|
|
|
|
|
|
|