MDX Bridge Results (Left Join)

StackOverflow https://stackoverflow.com/questions/23617017

  •  21-07-2023
  •  | 
  •  

Question

I'm new to data warehousing so I may be approaching this the wrong way and if so please let me know a better alternative. The following is an example using the same conceptual relationships but different names.

I have a dimension of faculty and I have a bridge (many to many) connecting those faculty to their specialties. A faculty member can have more than one specialty but they may have none. When I perform MDX queries and pull the specialty and facualty member the results are showing perfectly fine but I can't seem to figure out the best way to find faculty members that have no specialty and combine them together with the ones that do. Here is a quick snapshot of the results of a mdx query I want:

name    specialty       Salary (fact)
James   Biology         300
James   Bio-diversity   300
Henry   Mathmatics      350
George  NULL            100
Louis   Linguistics     240

etc...

This is what I'm getting from my current query:

name    specialty       Salary (fact) 
James   Biology         300
James   Bio-diversity   300
Henry   Mathmatics      350
Louis   Linguistics     240

If I take out the bridge relationship specialty then George shows up fine. Any help or suggestions?

Was it helpful?

Solution

I would add a member named e. g. "none" to the specialty dimension. Then I would add entries to the bridge table for all faculty members which have no specialty that reference this dimension entry.

To technically implement this in detail, there are several ways:

  • You can change your ETL process to add these records to the tables, which is the cleanest way.
  • You could also use views instead of the bridge and dimension table in your Data Source View, and the views contain the logic to add these records, probably using some kind of WHERE NOT EXITSTS logic.
  • And finally, instead of using views, you also could use named queries in the Data Source View instead of the views, but implementing the same logic, just on another level.

That last two implementations would have the advantage that the ETL process need not be changed.

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top