Hi all!

Could anyone please tell whether there is a way in AS to build cube based on
so called factless fact table when intersection of foreign keys is the fact
itself. If yes then any hints will be much appreciated.

Thanks in advance

can't believe nobody did it before since Kimball's tip #56 says:
"More advanced techniques like ... and factless fact tables all translate
smoothly"(into AS)
He must've had grounds to state that.

--

Thank you,

Alex
"Alex" <alex_remove_this_ (AT) telus (DOT) net> wrote

Alex,

I can't find a link to Kimball's Design Tip #56 (only tips to #50 (only
upto year 2003) are listed at:

http://www.kimballgroup.com/html/designtips.html


Anyway, it should be simple to build an AS cube for such a table:
arbitrarily choose any column to define a measure, but then change the
aggregation function of the measure from the default "Sum" (meaningless
for key columns) to "Count". This measure then counts the number of fact
table rows for any cube cell (assuming no rows are dropped by joining to
the relevant dimension tables).


- Deepak

*** Sent via Developersdex http://www.developersdex.com ***
Don't just participate in USENET...get rewarded for it!

Thank you very much Deepak,
as for tip 56 I got it via email the other day (I subscribed to getting the
tips)

Alex
"Deepak Puri" <deepak_puri (AT) progressive (DOT) com> wrote

Sure, you can do it. Just pick a fact table key (or a dimension table
column) as a measure -- typically you will use a count or distinct
count aggregation for the measure, since usually the point of factless
fact tables is for counting things.

"Alex" <alex_removethis_ (AT) healthmetrx (DOT) com> wrote