Set Analysis - Filter not applied

[alliedarmour]

Hello!

I'm relatively new to Qlik and I currently trying to get a grip on set expressions.

I have this simple expression:

Count(
{$<Leistungskatalog = {"CAT1"}, [Name] = {"*"}> * <FallBewegungKey = {"=Count(ICD) > 0"}>}
DISTINCT Fallnummer)

So in other words I want to get all records which match the CAT1 catalogue AND have a count > 0 of a specific measure while ignore the Name selection (I also tried to nullify by leaving the right side empty). 

If I remove the part with the count it's working but as soon as I add the intersection and I select a name, the filter is applied although it should not be. 

Count(
{$<Leistungskatalog = {"CAT1"}, [Name Arzt] = >}
DISTINCT Fallnummer)

This works for example. Any hints on what I am missing?

[marcus_sommer]

If you want to ignore the Name selection you need to apply this ignoring to all parts like:

Count(
{$<Leistungskatalog = {"CAT1"}, [Name]> * <FallBewegungKey = {"=Count(ICD) > 0"}, [Name]>}
DISTINCT Fallnummer)

or more simple:

Count(
{$<Leistungskatalog = {"CAT1"}, [Name], FallBewegungKey = {"=Count(ICD) > 0"}, [Name]>}
DISTINCT Fallnummer)

or in a more logical way like:

if(Count({< [Name] >} ICD) > 0,
Count({$<Leistungskatalog = {"CAT1"}, [Name]>} DISTINCT Fallnummer))

or in further optimized way like:

sign(Count({< [Name] >} ICD)) *
Count({$<Leistungskatalog = {"CAT1"}, [Name]>} DISTINCT Fallnummer))

Beside of this you might be also able to skip the count() part if no ICD records exists or they have the wrong respectively unwanted values - but this would depend on the dataset and the data-model if and how it might be possible.

[alliedarmour]

Hi Marcus,

first, thanks for your answer - ignoring the name in all selections makes sense, so I added the name filter to the second set aswell - but it still doesn't work.

But what works is if I also put the filter  inside the nested count like this:

Count(
{$<Leistungskatalog = {"CAT1"}, [Name] = > * <FallBewegungKey = {"=Count({$<Name = >} ICD) > 0"}, [Name] = >}
DISTINCT Fallnummer)

Is this normal / expected behaviour in Qlik Sense? 

[marcus_sommer]

Yes, this is the expected behaviour and it is related to the order of the execution and in which context they are performed. If by the count of ICD a certain selection should be ignored/overwritten this condition must be included in the inside of the expression and depending on view-requirements it might be included within all other stages of the calculation, too.