non comulative measures in IM - relational source

[jplarios]

How do you deal with non-comulatives in an IM with a relational source?

[Madelyn_SAS]

Can you be more specific?

[jplarios]

BIC_USER_3,
For example semi-additive summations. Kind of like a 'sum if' . Assuming my relation source has an architecture based on having all measures in a Fact table referencing dimension tables ( a dimensional approach).
One of this dimensions could be, say , time. Account balances and inventory levels are examples where you don't want to sum across all dimensions. For inventory levels, I would not want to average across all the dimensions. Having a function that restricts the measure would help.

This is easily done with an OLAP cube since it uses MDX. With a relational source, I would use SQL.

[jplarios]

Another example,
if a have 6 account balances that I select along with four periods how can I do an average , with a relational source, such that it would divide my results by the period cardinality of 4 instead of multiplying 4 * 6 ?
Again, I can see how an OLAP cube can achieve this via a calculated member but I can't figure it out in a relational source.