Calculation of metrics at database level or reporting level..? which is the best way..?


I just came across a same old situation of deciding where to do the calculation of certain metrics like percentages ?

Here is my take on it with a fictitious example –

Scenario 1:  Calculating it at database level and preserving it in the database:

see Fig.:1(Associate Level Sales) showing a table with sales made by a few sales executives in the years 2011 and 2012. Assuming that the percentage increase or decrease in the sales made by each of these executives is calculated at the database level and preserved in the database, the table appears as in the Fig.1 (Associate Level Sales).

Benifits: Now, since the calculations are materialized/preserved in the database, the performance of the reports containing this percentage calculation might be better, as the reports need to directly pull this calculation from the database instead of calculating on the fly.

Caveats: However in this scenario, the problems surfaces when these sales figures are expected to be aggregated at Manager level. See Fig.: 2(Manager Level Sales Pivot table) showing a pivot table built on top of the table data as in Fig.:1(Associate Level Sales) to arrive at the aggregated percentage change calculations. The percentage chanage calculations are wrong since they are doing the sum of the percentage changes at the associate levels to aggregate at the Manager level. whereas the calculations should have happenned as in the Fig.4.
In Fig.4 First manager level volumes are arrived at by adding up the associate level volumes. Percentage changes are then calculated on these manager level aggregated volumes.

PercentCalculationsDecisioning
PercentCalculationsDecisioning

Scenario 2:  Calculating it at the report level (e.g. pivot table in MS Excel or OBIEE or some other reporting tool) on the fly:

In this scenario, a pivot table is built as in scenario:1, but the percentag change calculation is done inside the pivot table as a calculated item using the formula (sales of 2012 – sales of 2011)/sales of 2011. see the results in Fig.: 3. Now these calculations match exactly with those of the Fig.: 4.

Benifits: Aggregations at higgher levels are calculated correctly.

Drawbacks: Since the percentage change calculation is not preserved in the database, it is calculated on the fly everytime the pivot table is refreshed resulting in report performance drop down.

Conclusion:  It’s just common sense that certain measures cannot be aggregated directly. Such measures are called Non-Additive measures. Additive measures should be preserved in database and Non-Additive measures should be calculated at the report level on the fly.

Hope you enjoyed reading this post. Please do share your comments, suggestions or questions, if any and help me improving this further..

Do rate this blog post on the top in terms of number of starts. The more you like it, the more the number of stars.. Happy time until my next blog post.. 🙂

Your valuable comments are highly honoured..!!

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Up ↑