Analyze dimension data without fact

3310714

Hi,

I have this simple model: 

D_REGION <--- F_SALES ---> D_STORES

If I want to get a list of regions (D_REGION) and stores (D_STORES), I get the error "Cannot find logical source coverage for logical columns"

If I include a column from F_SALES, then no issues.  Is there a way to force OBIEE to go through the F_SALES table when I'm just interested in analyzing dimension data? 

Gianni Ceresa

To do dimensional only analysis you can set the "Implicit Fact Column" in the subject area. This column will act exactly as you are asking, telling OBIEE which fact to implicitly use to join your dimensions without requiring you add it to your analysis.

[Deleted User]

+1 to Gianni but just to be more explicit. You seem to have a multi-fact model and not all fact tables tie to all dimensions. So setting an implicit fact column is just a part of the solution since when you use that you better pray that your model is correct and has the valid content levels set where needed.

3310714

I did that, but still getting the same error... 

3310714

I may have modeled the BMM incorrectly.  The diagram below is my physical diagram.  I have one Fact table in the middle and the rest are Dimension tables.  In my BMM, I have combined #2 and #3 into the same Logical Table because they have similar type of information.  Should these have been separated into it's own Logical Table, but combined in the Presentation Layer instead?   Should my BMM have 1 Logical Fact and 5 Logical Tables? 

[Deleted User]

2 and 3 can share the same logical table but must not be in the same logical table source since otherwise obi tries to join fact -> 2 -> 3 instead of fact -> 2 and fact -> 3.

Nesting things inside a single lts means that you try to snowflake

3310714

If I put them into the same Logical Table, the "Implicit Fact Column" doesn't work.  When I segregate them out, then it works.

Gianni Ceresa

Similar type of information isn't the same type of information ...

You generally add in the same logical table sources which complete each other (partitions like current data in a table and historical in another one) or same data at different level of details or the "reduced" snowflake.

But otherwise by putting them in the same logical table it doesn't work as Christian said.

If you want the end users to find these data in the same presentation table it's fine, but from a modelling point of view they are different logical tables.

[Deleted User]

Then as Gianni implied your logical model is fundamentally flawed