Best way to deal with mixed dimensionality

[#1]

What is the recommended way to deal with mixed dimensionality in QV? Presuming I have a Location, and Transaction channel dimension. Now I have a combined "Facts" table that contains both, Plan, and Sales data. Sales data is slicible by Location, but not by Transaction channel, Plan data is slicible by Transaction Channel but not by the location. Next, I define two metrics, one for Plan and another one for Sales, but I don't want filtering selections of Location dimension to be affecting Plan metric, and I don't want Transaction Channel filtering selections to be affecting Sales.

I know that I can achieve desired results by using set analysis and negating any selection made for an inappropriate dimension and that's easy enough to achieve (if you only have to metrics with mixed dimensionality). What's concerning to me is that intuitively, I sense that there's a CPU load and cache footprint overhead associated with putting this logic inside of set analysis. Is there a better way of dealing with said scenario?

P.S. Example above is put together for demonstration purposes only, actual application we have uses tens of metrics with various dimensionlities.

Thanks for taking the time!

[hic]

See e.g. http://community.qlik.com/blogs/qlikviewdesignblog/2012/10/26/mixed-granularity

HIC

[#3]

Hi Henric!

Thanks for the link, I've actually read your article before and thought that it was pretty damn cool. With that being said I believe that we're dealing with two separate issues, unless I am failing to see a way to utilize your solution for mixed granularity to my problem of mixed dimensionality.

In my case I have two separate dimensions. Transaction Channel dimension is completely logically separate from Location dimension, so the problem is not of mixed granularity, but of mixed dimensionality. I'll try to re-state the problem

Our combined fact table has two fields:

LocationKey

TransactionChannelKey

ExtNetDollars

PlanNetDollars

Two Dimensions:

Location

Transaction Channel

We want the filtering selections of Location only be applicable when we use ExtNetDollars for our measures, we want selections of Transaction Channel to only be applicable when measure uses PlanNetDollars. Is there a way to model the data in such way that we would not have to resort to using set analysis similar to the one below:

Sum({$<LocationKey=,>} PlanNetDollars)

Sum({$<TransactionChannelKey=,>} PlanNetDollars)

Being new to QlikView I might be completely off base here, or might be missing the point you're trying to make. In this case, I am sorry. Could you elaborate a bit on the solution you're offering?

[hic]

You can still use generic keys for solving this problem. Just assign the <ALL> symbol to the appropriate records.

This way a selection in "Location" will not remove any PlanNetDollars. See more on http://community.qlik.com/docs/DOC-3451

HIC