Show records where team assignment = 1 - sql

Question

by changing the below TeamRecords = 1 to = another number finds the rows with the amount I change to, its only sometimes its counting one too many which is odd. When a new Incident is created it has a unique number and every time a new assignment is added from the Task table it adds another row of the IncidentNumber, so you could have duplicate Incident number rows which I've remove with the seq = 1 below. When a new assignment is created it creates a new CreateddateTime in the Task table so for example you could do a Max(t.[CreatedDateTime] to find the last assignment of any IncidentNumber. So, the TeamRecords = 1 is what I need to find all records for that specific team where there is only 1 assignment for that team.
Does that help any?
Here is what I have so far...
Use TEST
Go
WITH RankResult AS
(
SELECT i.[IncidentNumber],
i.[CreatedDateTime],
i.[ResolutionDateAndTime],
i.[Priority],
i.[Status],
i.[ClientName],
i.[ClientSite],
t.[OwnerTeam],
t.[Owner],
row_number() over( partition by i.RecID
order by t.CreatedDateTime desc, t.OwnerTeam ) seq,
TeamRecords = COUNT(*) OVER(PARTITION BY t.ParentLink_RecID)
FROM Incident as i
Inner JOIN Task as t
ON i.RecID = t.ParentLink_RecID
WHERE t.OwnerTeam = 'Infrastructure Services'
AND i.CreatedDateTime >= '20121001'
AND i.CreatedDateTime <= '20131001'
)
SELECT DISTINCT
[IncidentNumber],
[CreatedDateTime],
[ResolutionDateAndTime],
[Priority],
[Status],
[ClientName],
[ClientSite],
[OwnerTeam],
[Owner]
FROM RankResult
Where TeamRecords = 1
And Seq = 1
Order By IncidentNumber Asc
GO

Answer 1

Using ROW_NUMBER means you will return the first assignment per team, not necessarily the teams with only one assignment. To do this you can use COUNT(*) OVER():
WITH RankResult AS
(
SELECT i.[IncidentNumber],
i.[CreatedDateTime],
i.[ResolutionDateAndTime],
i.[Priority],
i.[Status],
i.[ClientName],
i.[ClientSite],
t.[OwnerTeam],
t.[Owner],
TeamRecords = COUNT(*) OVER(PARTITION BY i.RecID)
FROM Incident as i
INNER JOIN Task as t
ON i.RecID = t.ParentLink_RecID
WHERE t.OwnerTeam = 'Info Services'
AND i.CreatedDateTime >= '20121001'
AND i.CreatedDateTime <= '20131001'
)
SELECT DISTINCT
[IncidentNumber],
[CreatedDateTime],
[ResolutionDateAndTime],
[Priority],
[Status],
[ClientName],
[ClientSite],
[OwnerTeam],
[Owner]
FROM RankResult
WHERE TeamRecords = 1;
2 things to note that I have changed in addition to the analytic function. Firstly I have changed your dates to the culture independant format yyyyMMdd, yyyy-MM-dd can still be ambiguous, so 2013-01-02 could be the 1st Feb or the 2nd Jan depending on your server/session settings. Secondly, Your where cluase was turning your join into an INNER JOIN anyway, so I just made it an INNER JOIN:
FROM Incident as i
LEFT JOIN Task as t
ON i.RecID = t.ParentLink_RecID
WHERE t.OwnerTeam = 'Info Services'
AND i.CreatedDateTime >= '20121001'
Here, if there is no match in Task then OwnerTeam will be NULL and NULL = 'Info Services' evaluates to false, so you will never return any rows with no match in Task thus making it an INNER JOIN). If you did in fact want a LEFT JOIN then you need to move this clause to the JOIN:
FROM Incident as i
LEFT JOIN Task as t
ON i.RecID = t.ParentLink_RecID
AND t.OwnerTeam = 'Info Services'
WHERE i.CreatedDateTime >= '20121001'

Answer 2

You can always use a simple subquery which groups across the relevant columns, counts them, filters where the count is 1, then joins it back to the main table to select the appropriate rows:
SELECT Incident.*
FROM (
SELECT OwnerTeam
FROM Incident AS Inc1
GROUP BY OwnerTeam
HAVING COUNT( * ) = 1
) AS Team
, Incident
WHERE Incident.OwnerTeam = Team.OwnerTeam
(Without more information, though, it's difficult to say if this will work for you.)