Question

I have a stored procedure that returns a set of rows for each ActivityUnitId.

What is needed is the Top 10 rows for each ActivityUnitId based on Hours. I have managed to achieve this, my query looks like

With Data AS
(SELECT  
   ProjectID,ActivityUnitID,Description,GroupID,
   SUM(Hours) AS NoiseHours,
   ROW_NUMBER() OVER(PARTITION BY ActivityUnitID ORDER BY SUM(Hours) DESC) as 'RowNum'
FROM         
   tbl_Sub 
INNER JOIN
   tbl_AnalysisData ON tbl_Sub.SubActivityID = tbl_AnalysisData.SubActivityID 
     INNER JOIN
       tbl_Analysis ON tbl_AnalysisData.LookupID = tbl_Analysis.LookupID
         INNER JOIN
           tbl_ActivityUnit ON tbl_ActivityUnit.ActivityUnitID = tbl_Sub.ActivityUnitID
           tbl_Suby.ProjectID = @ProjectID 
         AND
           tbl_Sub.ActivityUnitID = ISNULL(@ActivityUnitID,tbl_Sub.ActivityUnitID)
GROUP BY 
  ActivityUnitID,ProjectID,  Description,AUGroupID
) SELECT * from Data where RowNum<=10

The ‘RowNum’ column contains Row number for each row, value of which has been assigned based to hours. So the First 10 rows contain the rows with top 10 hours.

Now what I want is to add an Extra summary Row in the end for each ActivityUnitID. This will contain the sum of Hours for all the rows that have been left out i.e. summary Row for
RowNum > 10

So what I’ll have in the end is Top 10 rows for each ActivityUnitID + an extra row that would sum up the hours for other rows for that ActivityUnitID

For example lets say I have a table with 2 cols

ID  Hours     RowNum
1A    30         1
2B    20         2
3C    10         3
4D     5         4
5E     4         5
6F     3         6

How do I do a select on this so that I get Rows where RowNum <=3 and another row With summation of others

ID  Hours  
1A   30
2B   20
3C   10
Oth  12

Answer 1

The simplest way would be to UNION those totals with the data you already have

SELECT  NULL, NULL, 'Totals', NULL, SUM(NoiseHours), 11 AS RowNumber
FROM    Data 
WHERE   RowNum > 10

Your complete statement could then look something like this

With Data AS
(
SELECT  ProjectID,ActivityUnitID,Description,GroupID,
        SUM(Hours) AS NoiseHours,
        ROW_NUMBER() OVER(PARTITION BY ActivityUnitID ORDER BY SUM(Hours) DESC) as 'RowNum'
FROM         tbl_Sub 
            INNER JOIN
            tbl_AnalysisData ON tbl_Sub.SubActivityID = tbl_AnalysisData.SubActivityID 
            INNER JOIN
            tbl_Analysis ON tbl_AnalysisData.LookupID = tbl_Analysis.LookupID
            INNER JOIN
            tbl_ActivityUnit ON tbl_ActivityUnit.ActivityUnitID = tbl_Sub.ActivityUnitID
tbl_Suby.ProjectID = @ProjectID 
AND
tbl_Sub.ActivityUnitID = ISNULL(@ActivityUnitID,tbl_Sub.ActivityUnitID)
GROUP BY ActivityUnitID,ProjectID,  Description,AUGroupID
)
SELECT * from Data where RowNum<=10
UNION ALL 
SELECT  NULL, NULL, 'Totals', NULL, SUM(NoiseHours), 11 AS RowNumber
FROM    Data 
WHERE   RowNum > 10

The difficulty when using a union is that both parts of the union should have the same amount and same type of columns, hence the need for adding dummy columns to the select.