SAS OLAP Questions & Feasibility

[mcollins1264]

Hello,

I am new to using SAS OLAP. I have been using SSAS for several years. I just have a few comparison questions that I am curious about:

Is it possible to have multi-fact cubes, or to link cubes together? I.E., can I make calculations that refering to measures of different granularity, like comparing customer counts to demographic houshold counts and so forth.

Also, is it posslbe to somehow do many-to-many relationships, I.E. looking at counts of auto insurance policies with certain driver charateristics. This would be like counting if x attribute exists on a lower grained dimension, count it, otherwise do not ect.

[LinusH · Accepted solution]

1) I'm not sure how you want this to work Perhaps an example will clear this question.

There are not possible to link cubes together with different setup of metadata. Dimensions can be shared between cubes though. And there are some possibilities to have data stored on different levels using ROLAP structure.

2) Not sure about this one either, example? Something that might are similar is the NUNIQUE statistic. Or this could perhaps be solved by creating some MDX-expressions.

/Linus 

[mcollins1264]

Linus,

To your second point, I did some more investigating, and I think I NUNIQUE will work for the time being. M2M is neccessary in certain situations where you need other measures to be view this way (other than counts), but I can roll with this for now. I'm starting to understand that SSAS is very unique in this regard, as well as with the multi fact setup. For what I need, I can just run proc sql statements to pull from different cubes and then join them together, although that is not as easy on the end user. Thanks for your help--