Conformed dimensions

[matrixfree]

Hello,

I read on Conformed dimensions and found an example of when we need to creat e a CONFORMED Dimension even thought I still find this topic a bit difficult.

Can someone share a couple of examples where the need to create Conformed dimensions occurs.

[navissar]

I like the Kimball methodology a lot, but the use of lingo definitely obfuscates some of the great ideas there.

A conformed dimension is simply a dimension which is exactly the same across the business. This sounds very basic, and it is, but a lot of times we find that - mostly due to time pressure - dimensional data isn't conformed. For example, in one place I did a project for the security team needed to see all employees and contractors and their entry clearances etc. HR, on the other hand, was only interested in employees, without contractors; and senior management wanted their data to exclude temps. Since their dashboards were built by different contractors in different times, they had three employee dimensions to cater for this situation, and these dimensions weren't conformed - the each referred to a different population definition, had different fields with different labels (A contractor or a temp wouldn't have an employee ID, for instance) and so on.
I created a single employee dimension, containing all employees with all relevant data and a flag for temps and contractors, allowing us to create a perfect mathematical subset of the employees for HR and sr. management purposes.
Conformed dimensions are the corner stone of good data modelling because they allow us a single version of the truth - whenever you use the dimension to slice and dice your results will mean the same thing.

In FM modelling you could say that if you bring a data base query subject in once, and use it as the source of multiple model query subjects which form perfect subsets of it, then that is a conformed dimension.

[bdbits]

Employees is a good one (and nice write-up Nimrod). Customer or Vendor are very common, as are organizational dimensions (Company, Department, etc).

Another one that nearly everyone uses as a conformed dimension without thinking about it is the time dimension.

To me, a big advantage of conformed dimensions is that provides a consistent set of data and capabilities to your users. Regardless of which facts they are dealing with, they will come to know that if the Employee dimension is available, intuitively they know the attributes they will be able to use for slicing, dicing, and reporting in general.

[matrixfree]

Thank you all.

Question:
Who decides what the conformed dimensions will be ? The FM modeler or the functional Analyst or the DBA or a mix of ?

- are the "Common" DIMENSIONS the same as Conformed ?