Measures and Measure Groups - Expert Cube Development with SSAS Multidimensional Models - page 91

4. Define how each account type should aggregate. This can be done by
right-clicking on the name of the project in the Solution Explorer in
SSDT and selecting Edit Database . In the following screenshot, in the
Account Type Mapping table, the Name column contains the name of
the built-in or user-defined account types and the Alias column contains
a comma-delimited list of member names on the Account Type hierarchy
that map to each account type, and Aggregation Function allows you
to select either Sum or one of the semi-additive aggregation types for each
account type.
Most of this work can also be done by running the Define Account Intelligence
wizard. To do this, go to Dimension | Add Business Intelligence | Define
Account Intelligence .
In our experience, the ByAccount aggregation type is very rarely
used. This is partly because cube developers do not know about it,
and partly because most financial applications need to implement
more complex logic for aggregation than the built-in semi-additive
aggregation types allow. This means that aggregation logic is usually
implemented using MDX Script assignments, which give complete
flexibility although at the expense of a lot of complexity and do not
perform as well as the built-in aggregation types. Probably, the best
compromise is to start off by using ByAccount and then for the
accounts which need more complex aggregation, configure them to use
Sum in the Edit Database screen and then use custom MDX for them.