Hi,


Their are multiple type of Time Dimensions.
We have the default

Calendar Time Dimension
Then

Fiscal Calendar Time Dimension
Relative Time Dimension
Serial Time Dimension

Question can a measure be associated to multiple time dimensions?
Can you break down a measure by Calendar Time Dimension and then break down by a Serial Time Dimension when drilling down in the measure?

Would it be wiser to create a secondary measure and use the Serial Time Dimension in a seperate measure all together?

your feedback would be great.

Merci en avance.

Can you describe for us what these two things mean?

1) Serial Time Dimension?
2) Relative Time Dimension?

We've never heard of these before - and we spend a LOT of time on this subject. We sense these ideas might be internal vocabulary at your customer, so a good description will help us help you.

You can create more than one Dimension in PMF that you can relate to your consumers as being Time, but there is only one magically endowed Time Dimension in the system. Any additional Dimensions would not "know" they were Time Dimensions so there would be no time animation on Dashboards, or other related magic. Likewise, linkages in Measures would not be able to relate back to real Dates within the logic (although you could create custom logic that would do those relationships for loading purposes). Lastly, you would not be able to create Time ranges for these additional Dimensions that represent something about Time. You would have no sense of "prior" or "previous" periods.

But you could have all the aggregative logic that is inherent to Dimensions.

And, because Time Dimensions come with a whole set of responsibility, not having a Dimension be a "real" Time Dimension could be very liberating.

You could for example have dimensional sparsity (e.g., entire parts of Time that were non-pertinent could be missing from the available level values) which would not be possible with a "real" Time Dimension because there is no responsibility for contiguity. (Time Dimensions must be contiguous). You would not have to worry about the previous- or prior-ness of any Dimensional node. Etc.

thanks

Thanks for your answer , it does clear a lot of things up.

Here is the definition I have,
Serial Time Dimension (starts at a given point in time 2004 and starts counting forward from week 0, week 1, week 2, week 3 and day 0, day 1, day 2, day 3 and so on)

Relative Time Dimension (starts at a given point in time 2009 and counts down week -1, week -2, week -3 or count up from the given point in time 2009 week 1, week 2, week 3 and so on)

Regards