Skip to main content
Announcements
Join us at Qlik Connect for 3 magical days of learning, networking,and inspiration! REGISTER TODAY and save!
cancel
Showing results for 
Search instead for 
Did you mean: 
Anonymous
Not applicable

Linking tables without association

Hi all,

I am trying to link two table together while on purpose not having them join/associate.

Let me explain based on example below. I have two table (Table 1 and Table 2). Both have a project column with the same values in them. So obviously, I could have both tables join/associate in Qlikview and link them up. I have a specific reason that I don't want to do that so in order to avoid this to happen, I qualified Table 2 in the loading script.

Now that that is done, I do would like the hours from Table 2 to appear in Table 1. I was thinking I can maybe do this by capturing the project letter of Table 1 and use that in a set analysis to sum the hours but not sure how to do that. I tried this expression in Table 1 but it is not working:

=sum({<Table2.Project = {'Project'}>} Table2.Hours)

Any ideas? Many thanks in advance.

2018-10-10_9-49-15.png

10 Replies
swuehl
MVP
MVP

Maybe like

=sum({<Table2.Project = p(Project) >} Table2.Hours)


But this will not group the Table2.Project values if used in a chart with dimension Project.


You would need something like


=Sum( If(Table2.Project = Project, Table2.Hours))


This will create a cartesian join on runtime.

Anonymous
Not applicable
Author

Hi Stefan,

This works. Many, many thanks!

Danny

Anonymous
Not applicable
Author

Stefan,

One more quick question; I used your suggested formula to come to a solution:

=Sum( If(Table2.Project = Project, Table2.Hours))


However, I actually have two criteria to fulfil; the first one is Table2.Project = Project but I also have a second where Table2.Phase = Phase.


I adapted your formula to read like the one below, but rather than creating a smaller subset, it creates a much bigger one. Any thoughts?


=Sum( If(Table2.Project = Project, if(Table2.Phase = Phase, Table2.Hours)))


Thanks,

Danny

swuehl
MVP
MVP

Well, not 100% sure. I usually avoid using these kind of unlinked tables.

Try maybe

=Sum( If(Table2.Project & '-' & Table2.Phase = Project &'-'& Phase, Table2.Hours))


Or


Sum(

Aggr(

     If (Table2.Project = Project and Table2.Phase = Phase, Table2.Hours),

Table2.Project, Table2.Phase, Project, Phase)

)


If one of these work, you can add set analysis to limit the possible combinations of the run time JOIN:


{<Table2.Project = p(Project), Table2.Phase = p(Phase) >}

Anonymous
Not applicable
Author

Thanks Stefan. but neither of the options worked. This is what I have (I used the real fields so I don't have to keep converting).

Any suggestions are welcome, in the meantime I'll keep looking.

=sum({$<[BUDGET.Expense Category]={'Labor - Client Service'}, BUDGET.PID = p(PID), [BUDGET.Work Phase] = p(Phase)>}  if(BUDGET.PID&'-'&[BUDGET.Work Phase] = PID&'-'&Phase, BUDGET.FY19))

Thanks!

swuehl
MVP
MVP

"Work Phase" and "Phase"  show the same values?

Can you upload a simplified sample QVF or some screenshots?

Also a view of your data model (to see the tables involved) may help.

Anonymous
Not applicable
Author

Stefan,

Please find attached. Let me know if something else may be more helpful.

2018-10-11_8-52-45.png

2018-10-11_8-51-53.png

swuehl
MVP
MVP

Ok, Project and Phase are part of a synthetic key and are located in two to three resident tables in your model.

I've seen some issues with large synthetic keys (and I am not sure what the run time cartesian JOIN will do in such a case).

Can't you concatenate your actual and forecast fact table to eliminate the huge syn key table?

Also, remove Month and Year from the fact table, creating a Year-Month field in the script and link this to your financial year calendar.

Anonymous
Not applicable
Author

Hi Danny

You can use ApplyMap to and could avoid this kind of issue

https://help.qlik.com/en-US/sense/June2018/Subsystems/Hub/Content/Scripting/MappingFunctions/ApplyMa...

Regards,