About Levels of Aggregation

Only perform aggregations on measure columns. Measure columns should exist only in logical fact tables.

You can select different aggregation rules for different dimensions that are associated with a logical column. Suppose someone queries the aggregate column along with one dimension, you may want to use one type of aggregation rule, whereas with another dimension, you may want to use a different aggregation rule. For example, number of employees is a count on all dimensions except on the time dimension where the aggregation rule would be last.

When the default aggregation rule is Count Distinct, you can specify an override aggregation expression for specific logical table sources. For example, you may want to specify override aggregation expressions when you're querying different logical table sources that already contain some level of aggregation.

You can choose the EVALUATE_AGGR aggregation rule to enable queries to call custom functions in the data source. Use this aggregation rule when the aggregation must be done in an external data source.

By default, data is considered sparse. However, you might have a logical table source with dense data. A logical table source is considered to have dense data if it has a row for every combination of its associated dimension levels. When setting up aggregate rules for a measure column, you can specify that data is dense only if all the logical table sources to which it's mapped are dense.

For measures where the aggregation rule is the same in all dimensions, select one of the aggregate functions from the Aggregation Rule list. The function you select is always applied when a user or an application requests the column in a query, unless an override aggregation expression has been specified. When you select Count Distinct as the default aggregation rule, you can specify an override aggregation expression for specific logical table sources. Choose this option when you have more than one logical table source mapped to a logical column and you want to apply a different aggregation rule to each source.