SQL Reporting Aggregates and Joining Issues - sql

Question

I am attempting to create a report for orders moving through the various actions on a weekly basis. Could you guys help me out with it?
The desired report (simple ...right):
Week Of Action 1 Action 2 Action 3
---------- -------- -------- --------
2011-11-07 34 55 35
2011-11-14 34 55 35
And here is what I have tried so far, all of which have been miserable failures:
CREATE TABLE WorkOrderHistory
(
WorkOrderHistoryID int, --PK
WorkOrderActionID int,
DateCompleted datetime
)
CREATE TABLE WorkOrderAction
(
WorkOrderActionID int --PK
)
DECLARE
#StartDate DateTime,
#EndDate DateTime,
#SeasonID int;
SET #SeasonID = 16;
SELECT #StartDate = '2011-09-01',
#EndDate = ISNULL((SELECT TOP 1 StartDate FROM Season WHERE SeasonID > #SeasonID), GETDATE()) -- End date will be set to the current date if no season exists beyond the current
--Method 3 Inner Joins. This fails because of my attempt to join on the WeekOf alias to the DATEADD function
SELECT DATEADD(WEEK, DATEDIFF(WEEK, '19000101', WOH.DateCompleted), '19000101') AS WeekOf,
ArtworkCapture.WOsProcessed AS ArtworkCapture
FROM WorkOrderHistory WOH (NOLOCK)
INNER JOIN WorkOrderAction WOA (NOLOCK) ON WOH.WorkOrderActionID = WOA.WorkOrderActionID
INNER JOIN (SELECT COUNT (*) AS WOsProcessed,
DATEADD(WEEK, DATEDIFF(WEEK, '19000101', WOH.DateCompleted), '19000101') AS WeekOf
FROM WorkOrderHistory WOH (NOLOCK) INNER JOIN WorkOrderAction WOA (NOLOCK) ON WOH.WorkOrderActionID = WOA.WorkOrderActionID
WHERE WOH.DateCompleted >= #StartDate AND WOH.DateCompleted < #EndDate
AND WOH.WorkOrderActionID = 1 --Artwork Capture
GROUP BY DATEADD(WEEK, DATEDIFF(WEEK, '19000101', WOH.DateCompleted), '19000101')) ArtworkCapture ON WOH.WeekOf = ArtworkCapture.WeekOf
WHERE WOH.DateCompleted >= #StartDate AND WOH.DateCompleted < #EndDate
GROUP BY DATEADD(WEEK, DATEDIFF(WEEK, '19000101', WOH.DateCompleted), '19000101')
ORDER BY DATEADD(WEEK, DATEDIFF(WEEK, '19000101', WOH.DateCompleted), '19000101')
--Method 2 Subqueries. I can not figure out how to properly form this query.
SELECT DATEADD(WEEK, DATEDIFF(WEEK, '19000101', WOH.DateCompleted), '19000101') AS WeekOf,
(SELECT COUNT (*)
FROM WorkOrderHistory WOH (NOLOCK) INNER JOIN WorkOrderAction WOA (NOLOCK) ON WOH.WorkOrderActionID = WOA.WorkOrderActionID
WHERE WOH.DateCompleted >= #StartDate AND WOH.DateCompleted < #EndDate
AND WOH.WorkOrderActionID = 1 --Artwork Capture
GROUP BY DATEADD(WEEK, DATEDIFF(WEEK, '19000101', WOH.DateCompleted), '19000101'))
AS ArtworkCapture,
(SELECT COUNT (*)
FROM WorkOrderHistory WOH (NOLOCK) INNER JOIN WorkOrderAction WOA (NOLOCK) ON WOH.WorkOrderActionID = WOA.WorkOrderActionID
WHERE WOH.DateCompleted >= #StartDate AND WOH.DateCompleted < #EndDate
AND WOH.WorkOrderActionID = 3 --Art Entry
GROUP BY DATEADD(WEEK, DATEDIFF(WEEK, '19000101', WOH.DateCompleted), '19000101'))
AS ArtEntry
FROM WorkOrderHistory WOH (NOLOCK) INNER JOIN WorkOrderAction WOA (NOLOCK) ON WOH.WorkOrderActionID = WOA.WorkOrderActionID
WHERE WOH.DateCompleted >= #StartDate AND WOH.DateCompleted < #EndDate
GROUP BY DATEADD(WEEK, DATEDIFF(WEEK, '19000101', WOH.DateCompleted), '19000101')
ORDER BY DATEADD(WEEK, DATEDIFF(WEEK, '19000101', WOH.DateCompleted), '19000101')
--This query gives me all of the data I need but it is not aggregated, so there is a record for each action per week so [2011-11-07 - 1 - 34], [2011-11-14 - 1 - 34], [2011-11-07 - 2 - 55], [2011-11-14 - 1 - 55].
SELECT DATEADD(WEEK, DATEDIFF(WEEK, '19000101', WOH.DateCompleted), '19000101') AS WeekOf,
WOA.WorkOrderActionID, COUNT (*) AS WorkOrdersProcessed
FROM WorkOrderHistory WOH (NOLOCK) INNER JOIN WorkOrderAction WOA (NOLOCK) ON WOH.WorkOrderActionID = WOA.WorkOrderActionID
WHERE WOH.DateCompleted >= #StartDate AND WOH.DateCompleted < #EndDate
GROUP BY DATEADD(WEEK, DATEDIFF(WEEK, '19000101', WOH.DateCompleted), '19000101'), WOA.WorkOrderActionID
ORDER BY DATEADD(WEEK, DATEDIFF(WEEK, '19000101', WOH.DateCompleted), '19000101'), WOA.WorkOrderActionID

Answer 1

Looks Like you want to convert rows into columns. You don't need to use sub queries use either Pivot (for SQL 2005+) or SUM/CASE.
Notes: I didn't bother with the Week Calculation since you already have that. I also added a name for WorkOrderAction, since I assume you have that rather than just an ID
Declare #WorkOrderHistory as table
(
WorkOrderHistoryID int, --PK
WorkOrderActionID int,
DateCompleted datetime
)
declare #WorkOrderAction as TABLE
(
WorkOrderActionID int, --PK
Name varchar (50)
)
Insert into #WorkOrderAction
Values (1, 'Action1'),
(2, 'Action2'),
(3, 'Action3')
INSERT INTO #WorkOrderHistory
VALUES (1, 1, '2011-11-07'),
(2, 1, '2011-11-07'),
(3, 1, '2011-11-07'),
(4, 2, '2011-11-07'),
(5, 3, '2011-11-07'),
(6, 3, '2011-11-07'),
(8, 2, '2011-11-14'),
(9, 2, '2011-11-14'),
(10, 2, '2011-11-14'),
(11, 3, '2011-11-14'),
(12, 3, '2011-11-14'),
(13, 1, '2011-11-14')
SELECT datecompleted,
action1,
action2,
action3
FROM (SELECT woa.name,
COUNT(woh.workorderhistoryid) AS kount,
datecompleted
FROM #WorkOrderHistory woh
INNER JOIN #workOrderAction woa
ON woh.workorderactionid = woa.workorderactionid
GROUP BY woa.name,
datecompleted ) p PIVOT ( SUM(kount) FOR p.name IN (
[Action1],
[Action2], [Action3] ) ) AS pvt ​
Produces this result
datecompleted action1 action2 action3
------------------ ------- ------- -------
2011-11-07 0:00:00 3 1 2
2011-11-14 0:00:00 1 3 2
see sample at data exchange