Author Topic: Transformer Cube With Multiple Fact Table and Dimension Related to One Fact  (Read 1325 times)

Offline CognosGeek

  • Associate
  • **
  • Join Date: Dec 2016
  • Posts: 2
  • Forum Citizenship: +0/-0

I have to provide a powercube which can be analysed by user in Analysis Studio.

Now I have two fact tables and some dimensions. I have created a FM package and using that as a source in Transformer cube.

Now I have one dimension specific to one of the Fact table. I have included that also. After building the cube and analyzing it in Analysis studio the measures data is not showing correct records when I include that dimension.

This dimension is needed by my client to compare the measures from both fact table but since it is related with only one fact table . I am not sure how to design it in transformer or whether it can't be achieved in transformer.

Any help or leads would be appreciated to resolve this issue.

Offline bdbits

  • Super Moderator
  • Statesman
  • ******
  • Join Date: Feb 2010
  • Posts: 1,820
  • Forum Citizenship: +106/-0
So the client wants to use this one dimension but it has no relationship to the fact table? That really does not make any sense - why would they do that?

You can have multiple fact tables in a single cube, and not all dimensions have to be related. But you need to only use the dimensions associated with a fact table when selecting those measures. With no relationship, it's a bit like doing a cross-join in SQL.

Going out on a limb, you could add a member for 'unknown' to this dimension, then add the key for that member to the unrelated fact table. Whether this will resolve your particular issue is difficult without knowing the semantics of your data.