Do not input private or sensitive data. View Qlik Privacy & Cookie Policy.
Skip to main content

Announcements
Qlik GA: Multivariate Time Series in Qlik Predict: Get Details
cancel
Showing results for 
Search instead for 
Did you mean: 
bilzekek
Contributor III
Contributor III

Compose DWH modelling question

Hi,

How would below modelling in Compose act.

I have 3 tables/entities in my DWH layer, that should be joined to 1 dimension in the Data mart. The issue is that Table A (the main table) has a relationship to Table B on a column that is not the Key in Table B. Tried to picture it below.

I was wondering how you would design this - and if design below is used, how Compose will handle multiple rows in Table B with same value in the column use for the relationship;

Untitled.png

Labels (2)
1 Solution

Accepted Solutions
bilzekek
Contributor III
Contributor III
Author

Hi,

Yes that is the issue and all tables should be part of 1 dimension - not fact. I was able to "fix" this, by creating a "translate" table between Table A and Table B, that only contained A_ID and B_ID, but where I made A_ID the key. Table A then relates to this new table, and the new table related to Table B.

Dont know if this is the best way to fix it though.

View solution in original post

2 Replies
BHR
Contributor III
Contributor III

I believe it will all be based on what you define as the root of your dimension in the data mart. Anything that is a "parent" of the root of the dimension should be available to include in the dimension but anything that is a "child" will not. In the case above - if you choose table A as the root for your dimension then I don't believe that table B would be available for inclusion.

You might possibly be able to setup table B as a factless fact, or a join table fact. Table B would then become the Face in the middle of the star schema while Table A and Table C could be 2 dimensions that relate to the Fact table.

bilzekek
Contributor III
Contributor III
Author

Hi,

Yes that is the issue and all tables should be part of 1 dimension - not fact. I was able to "fix" this, by creating a "translate" table between Table A and Table B, that only contained A_ID and B_ID, but where I made A_ID the key. Table A then relates to this new table, and the new table related to Table B.

Dont know if this is the best way to fix it though.