Aggregation

Aggregation in the world of business intelligence is of considerable importance.

Usually, we're talking about aggregation tables with summarized data instead of detailed tables. As with many issues, aggregation has pros and cons.

Aggregation is a way to implement partial groups of multidimensional cubes in practice.

The main advantage of summarized tables is improved performance.

The main disadvantage is that more detailed data cannot be reached (unless you maintain both aggregated summary tables and detailed data, which then raises issues of location and consistency). Additionally, it requires maintenance.

When is it still worthwhile to use aggregation? (Some recommendations based on a SAP lecture):

• When the ratio between base and summary data is significant (compression of at least ten times).

• When dealing with popular accesses that are accessed more often, make sure not to aggregate on all dimensions, only on some...

• Not too focused, but not too generalized either. Look for the golden path.

• When dealing with data accessed recently (not just those useful a year ago).

Business aspects should also be considered:

To what extent will there be a need to access secondary data for decision-making?

If more detailed data is requested, what is the cost of delay when the data is unavailable?

On the other hand:

Does adding data add value for the user, or is it more confusing and not contributing to decision-making?

As mentioned, there are no definitive answers, but there are certainly considerations here that will help you decide.