Unlock a world of possibilities! Login now and discover the exclusive benefits awaiting you.
Hi,
I'm struggling with creating a model for tables that have Part of the mutual field relative to one another.
For Example I have 3 different tables with these fields:
Table1, fields:
A,
B,
C,
D,
E
Table2, fields:
A,
B,
F
Table3, fields:
A,
C,
H
In SQL for example if I would want to join Tbl2 and Tbl3 to Tbl1 I would left join tbl2 on fields A and B and left join tbl3 on fields on A and C to generate 1 big table having all the columns.
But I cant figure out how to do the same in Model in QlikSense. I need to be able to use all of the columns later on in the Dashboard Creation step.
How Can I create a correct connection between the 3 tables using KeyTable? Or maybe something else?
Thank you 🙂
In general an appropriate link-table containing the distinct (combined) keys of all associated tables be created. But it won't be a benefit - neither from the needed efforts and complexity nor in regard to the performance - much more likely would be a big disadvantage.
The officially recommended data-model is a star-scheme with a single fact-table and n dimension-tables. And the bigger the data-sets are the more important is this recommendation.
Hi @Jenny ,
You can try creating a composite key using two columns to join the tables. For example:
- To join Table1 and Table2, create a key by combining columns A and B (e.g., A&B as Key1) in both tables.
- Similarly, to join Table1 and Table3, create a key using A and C (e.g., A&C as Key2) in both tables.
Joins in Qlik aren't completely equally to those in sql but the essential logic is the same and therefore what you are describing should be working in Qlik, too.
t: load * from t1; left join(t) load * from t2; left join(t) load * from t3;
But I dont want to join them straightforward Like I would do in SQL because my real tables are really big so I dont want to create 1 huge single table.
Is there a way using KeyTable here?
In general an appropriate link-table containing the distinct (combined) keys of all associated tables be created. But it won't be a benefit - neither from the needed efforts and complexity nor in regard to the performance - much more likely would be a big disadvantage.
The officially recommended data-model is a star-scheme with a single fact-table and n dimension-tables. And the bigger the data-sets are the more important is this recommendation.
Thank you.