Archives of the TeradataForum

Message Posted: Fri, 19 Jul 2002 @ 14:18:38 GMT


	<Prev	Next>		<<First	<Prev	Next>	Last>>

Subj:		Re: Attributes at a lower level of granularity than metrics.

From:		John Grace

Jim,

This is a great question.

What we typically do is virtually allocate the metric (premium) to the atomic level and re-aggregate as required in all cases (your last option). You may also want to try multiple "fact" tables (virtual or physical) to perform unique premium allocations and aggregations depending on the users request.

For example, if the female/male premium allocations are different by region, one percentage won't fit all allocations. I guess another way to think about it is that the allocation percentage is your real atomic metric (fact table). All queries are targeted to select a sub- set of the allocation "fact table" which is product joined into the smaller "premium" table. This approach would allow very accurate allocation across many dimensions.

I hope this helps.

John Grace