Building a Time Dimension with Weekly Data

[salnick4]

I am fairly new to Transformer,a nd recently came back from a Cognos class, but I am experiencing some problems with Time Dimension.

My data is loaded weekly into the data warehouse, so when building my Time Dimension, I am dealing with weeks that are loaded every Saturday...so for example my weeks in Feb'07 would be 2007-02-03, 2007-02-10, 2007-02-17, 2007-02-17, 2007-02-24.

My issue is that after using the date wizard within Transformer to create the Time Dimension, and then generating the categories, I see extra weeks created for certain months.  For example in the month of Mar'07, I would see a week called 2007-03-01, and then a week called 2007-03-03.  The second would be the correct one, but for some reason I get this week with a few days in it. 

I believe that it has something to do with weeks that carry over to the next month.  I need to have Weeks, Months, Quarter, and Years in my cube.....with Weeks being the lowest level.  In fact the Weeks will be used as a dropdown within Cognos ReportNet reports...so it has to match what is in the data warehouse.

Please advise how I can build this dimension with the appropriate rool up to months and quarters without adding these additional weeks.

Thanks!

[anoop_ind2]

You should set up the appropriate values for the Partial Weeks option in the Dimension map in Transformer. I guess it is set to Always Split currently in ur case.

Jus set it to First Period if you want Mar 3 to be under February Month or set to Last period if you want Mar 3 under March.

[salnick4]

Thank you very much!!!

i will definitely try to use your suggestion.

;D