How to average/sum data in a day in SQL Server 2005

Question

I\'m trying to average data in SQL Server 2005 in a day. Here is what my database look like this if I use simple query as

SELECT timestamp, FEED
FROM ROASTE         


        

Answer 1

@marc_s hit upon the preferred solution of using a computed columns to handle this but to do it on-the-fly, cast your timestamp data to shear off the time component. In my example, I cast it to a 10 character field that contains the 0 padded date.

;
WITH ROASTER_FEED ([timestamp],[Feed]) AS
(
SELECT CAST('02/07/2011 12:00:01' AS datetime),1246   
UNION ALL SELECT '02/07/2011 12:00:01',1234    
UNION ALL SELECT '02/07/2011 12:00:01',1387    
UNION ALL SELECT '02/07/2011 12:00:02',1425   
UNION ALL SELECT '02/07/2011 12:00:03',1263   

UNION ALL SELECT '02/07/2011 11:00:01',1153    
UNION ALL SELECT '02/07/2011 11:00:01',1348    
UNION ALL SELECT '02/07/2011 11:00:01',1387    
UNION ALL SELECT '02/07/2011 11:00:02',1425    
UNION ALL SELECT '02/07/2011 11:00:03',1223    

UNION ALL SELECT '03/07/2011 12:00:01',1226    
UNION ALL SELECT '03/07/2011 12:00:01',1245    
UNION ALL SELECT '03/07/2011 12:00:01',1384    
UNION ALL SELECT '03/07/2011 12:00:02',1225    
UNION ALL SELECT '03/07/2011 12:00:03',1363
)
SELECT
    -- Shear off the time value by forcing the date into date values
    -- If this was SQL Server 2008+, we'd just cast to date type and
    -- be done with it
    -- Chart for convert options
    -- http://msdn.microsoft.com/en-us/library/ms187928.aspx
    CONVERT(char(10), RF.[timestamp], 121) AS [timestamp]
,   COUNT(1) AS row_counts
,   SUM(RF.Feed) as ValuesSum 
,   AVG(RF.Feed) as Average
FROM 
    ROASTER_FEED RF
GROUP BY
    CONVERT(char(10), RF.[timestamp], 121)    

Results in

timestamp   row_counts   ValuesSum  Average
2011-02-07  10           13091      1309
2011-03-07  5            6443       1288