How to load data - Uneven Heirarchies

[dssd]

I have time dimension with the hierarchy set as Year, Month. Data is loaded at the leaf level, which would mean at the month level. But, if i have multiple years in my time dimension there would be multiple instants of each month. For e.g jan would be there under both 2008 & 2009. How do we load data in such a case? Wouldnt it get loaded in the wrong way?

[nishant.thakar]

Hi dssd,
Can you please let us know where are you trying to load the data.

If you are taking about loading data into DB/DW then create a key which would be concat of "YEAR" and "MONTH" i.e. for Jan 2010 and Jan 2011 it would populate 012010 and 012011 respectively. You can then use this Key in your FM/Transformer/TM1 as a Key for generating Hierarchy.

Hope this helps.

Regards,
Nishant Thakar
"Never let your curiosity die"

[dssd]

Load the data into TM1. Do we typically have the leafs as Jan, Feb, March....etc or do we use ids like 201101. So that they are uniquely identifiable. This is a question relevant to TM1. I am aware of how it works in Transformer

[OLAPBPMguy]

If your month dimension rolls up into years and spans multiple years of format yyyy-mm, mmm-yyyy, yyyymm or similar then no element "Jan" or "Feb" etc actually exists.  Possibly there might be an attribute called "month name" or similar holding the string value of the month name.  For example element 201202 might have month name attribute of "Feb" but you can't load data to Feb you have to load data to 201202.

Think about it. This is the only way it can work with a linear time dimension.

You can find some good debate elsewhere on the various merits of TM1 cubes with single ("linear") time dimensions versus the alternative of splitting time into separate dimensions.  If Year and Month were separate dimensions then of course you could load to "Feb" but would have to specify a year coordinate as well.

TM1 is OLAP in the purest sense.  Measure and time dimensions are treated the same as any other dimension or fact.  There is no concept of group by for attributes (other than using attributes to build hierarchies within dimensions for aggregation, which TM1 does very well.)  To understand TM1 cubes you have to get out of the "relational mindset."

Do you actually have access to a TM1 server to play with? Sorry to ask but it seems that this and your last question would be easily answered by 5 minutes of practical experimentation.