Hitachi Consulting BI Blog

Defining ‘By Account’ Aggregations

In Yukon, you can use semi-additive functions such as Last NonEmpty Child to aggregate a measure. The function then works against the Time dimension. You can also identify a measure’s aggregation as “By Account”, which means that it uses the Account Type attribute for each member to choose which aggregation function to use. When you use the Wizard to build the dimension, you get to map your labels to the 7 pre-defined account types.

But what if you want to use a different aggregation function for a type? For example, what if you want to use “Average of Children” instead of “Last NonEmpty Child”? Or what if you change your mind about the Account Type mapping (and you really meant to spell it as “Asset” rather than as “Assset”)? The definitions of the account types are stored in the <xxx>.database file, which is the only XML definition file you cannot see by choosing View Code in the Solution Explorer. You need to go to Windows Explorer and open the file.

In practice, it’s probably best to try to use the bucket names as meaningfully as possible, but you don’t really have to. Think of the seven pre-defined account types as simply seven slots for aggregate functions. (There are 12 aggregate functions, so there are always 5 you won’t be able to use :-(. ) You could, if you want, assign seven different functions to seven “Account Types” and then map those to your account type names in the database file.

But whether you’re vanilla or chocolate fudge sundae in the way you use the Account Type mappings, the database file is the place to change them.

Reed