How to Model and/or query 'Similar' Sales

Hello,

I've been set an interesting task of trying to model in an SSAS 2012 cube a method of returning Sales data from a Measure group based on some Dimension attribute filters plus other some tolerance of closeness to the input filters.

For example:

If I had 1 Measure Group M, with 3 Dimensions, D1 D2 and D3 I might slice my Measure group by D1.attribute1, D2.Attribute1 and D3.Attribute.  

What I would then like to do is return all Sales rows matching the base filters but then look at other Sales rows where the attributes may be similar and return those.  

E.g.  D1.attribute1, D2.attribute1, and D3.attribute2 could be similar sales, whilst something like D1.attribute1111, D2.attribute1123, and D3.attribute5555 could be considered way out of distance.

Once i've got a set of exact and 'close' potential matches I can then use this data for further calculations in MDX.

What I am not sure of is:

1) Can I do this kind of thing purely in an SSAS model in the DSV with Meaures and calculations

2) Do I need some kind of mining model to generate some kind of cluster that I can query to return a base set of Sales that I can then further manipulate.

Performance is also a must!

Any suggestions are appreciated.

Regards,

MrHH


September 3rd, 2015 6:18pm

This topic is archived. No further replies will be accepted.

Other recent topics Other recent topics