OBIEE 11.1.1.7

Hyperion 6.5 User

Hi,

I'm having a problem with one of my facts. If I add another fact to my analysis. Aggregated data is wrong (reproduced) for this fact ...

I have recreated this fact in the repository, the problem persists ...

Thomas Dodds

What's the grain of these facts?  What's conformed?

Hyperion 6.5 User

Here is a view of the physical tables, they are identical.

In the business model and correspondence, here is the detail of the source table

Thomas Dodds

Is the content Levels (Contenu) the same for the DWH_STAT_FACT_ADMHOURG?

Hyperion 6.5 User

Yes the content levels are the same between the 2 tables. They point to a temporal hierarchy. And the granularity is the same ...

Robert Angel

Things to check; -

What is the key to the conformed dimension for both tables.

Do both facts return consistent results with the conformed dimension.

If you ran the equivalent physical sql of fact - dim - fact with the joins you have created do you get a Cartesian product.

- Generating the physical sql that is actually run and checking the joins may yield a big clue here!

And as @Thomas Dodds as advised, check the content levels on the logical table sources are correct and consistent on the two facts and the conformed dimension they are joining to.

And are the tables that do not have a content level set tables not joined to?

Hyperion 6.5 User

Hi Robert,

a few months later ... busy on other things...

Both facts return consistent data ... this error loses me ...

If I take the single fact "Admissions classic hospitalization" (the one that causes me problems) I can explore the data and I have no errors.

The problem arises when I add one of the other facts to an analysis.

These facts share the same granularity, the fact tables have identical fields, the joins at both the physical level and the logical level are identical ...

BUT

here is an error message that I had not seen ....

Thanks for your help

Robert Angel

Hi,

that would say to me that your column Admissions.Admission hospitalisation classique (sum) is not mapped for all of its sources, you need to open the definition for that column and ensure that it has underlying physical or logical mappings for all its sources.

Hope this makes sense, happy to give further clarification if not.

Hyperion 6.5 User

Hi Robert and thanks for that quick response !!!

Here is the physical diagram :

everything is an accordance...

Also correct at the logical level

I will deleted this fact and recreate it entirely ... can be ...

Robert Angel

Hi,

all of your logical table sources FACT => Dimension should have a content level set, typically at the lowest level of the dimension hierarchy.

To find this open up the folder under the business model table and click on it to see the underlying logical table source, the content level is therein.

Each dimensions physical table needs to have this set, usually to detail (the lowest level on its hierarchy).

Each FACT should have an entry for each dimension it is joined to, and each needs to again be set, usually to detail again.

I fear my French is not up to your diagram, but the Niveau Logique looks like what I am referring to and the blank spaces next to Dim Lieu Recep AppelDim and the two below it should not be there if the Fact table truly joins to all the dimensions.

This should be fixed, it could be part of your issue, but what I was originally asking you to check was that the measure value in the fact table that the error was pointing to was populated for all its data sources.

I am being generous with my explanation, I hope you will be generous with your point allocation  ;-)