CALCULATE - The Queen of all DAX functions: From Row context to Filter context

This post is an update on AVERAGEX with SUM vs. SUMX with CALCULATE(AVERAGE). With special thanks to Marco and Alberto explaining the topic of Evaluation Contexts so well! :)

A little while ago I wrote about the different outcome of 2 measures, when trying to find a solution for a problem. Back then I didn't know why they had a different outcome. Since I have been attending the SSAS BISM Tabular workshop by Marco and Alberto this week, I couldn't resist to look at the problem again.

Let's start with the 2 measures:

(1)
AVERAGEX(
        FILTER(
              Organisatie,
              Organisatie[Type] = "Woonlocatie"
        ),
        SUM(Productie_OrganisatieGroei[AantalClienten])
)

(2) 

SUMX(

    Productie_OrganisatieGroei,

    CALCULATE(

             AVERAGE( Productie_OrganisatieGroei[AantalClienten]),               

             Organisatie[Type] = "Woonlocatie"

    )

)

To understand the relationships between the tables, there is a 1-to-Many relation between Organisatie and Productie_OrganisatieGroei:

Datamodel

Starting with measure 1, AVERAGEX is an iterator, so it creates a Row context on Productie_OrganisatieGroei. The FILTER on Organisatie however, because it is in a Row context, is not propagated to the many-side. So calculating the SUM of [AantalClienten] doesn't take the filter into account and takes the sum of the whole Organisatie table.

Measure 2 also starts with an iterator (SUMX) which creates a Row context, but then uses CALCULATE inside, which creates a Filter context on that one row. Remember that Filter context does propagate over relationships. Now that we have created a Filter context by using CALCULATE, the filter on Organisatie[Type] does propagate through the relationship and filters the Productie_OrganisatieGroei table. That is why this formula does produce a correct answer on the problem I had.

Once more the pivot table which has measure 1 and 2 in it:

Note: although the project was done with PowerPivot v1, I just imported the xlsx-file in SQL Server Data Tools (Visual Studio / former BIDS) to extract the datamodel for my workbook:

AVERAGEX with SUM vs. SUMX with CALCULATE(AVERAGE)

I'm still working on the Management Information Dashboard I previously talked about when I was Implementing Parent / Child Hierarchy. (Note: There are some Dutch words and expressions in the examples)

Now I ran into this issue when trying to calculate the number of clients of a particular type of cost center (Activiteitencentrum, Ambulant, etc.).

My implemented hierarchy is in the Organisatie table, I'm interested in the organization type Woonlocatie and I want to calculate the number of clients per Region / Cost center. I have the Region and Type (both of table Organisatie) in the row label, where Year is in my column label.

I first tried this with Measure1:

AVERAGEX(FILTER(Organisatie, Organisatie[Type] = "Woonlocatie"), SUM(Productie_OrganisatieGroei[AantalClienten]))

But Measure1 does not calculate the totals per region correctly, it takes all the clients into account, not only where type="Woonlocatie". Then I tried Measure2:

SUMX(Productie_OrganisatieGroei, CALCULATE(AVERAGE(Productie_OrganisatieGroei[AantalClienten]), Organisatie[Type] = "Woonlocatie"))

which does take into account the organization type in the hierarchy. I'm not all clear about why this is working correctly with Measure2 and not with 1. If anyone does know why or has something to share about this, I'd like to know.

Measure1: AVERAGEX(FILTER(Organisatie, Organisatie[Type] = "Woonlocatie"), SUM(Productie_OrganisatieGroei[AantalClienten]))

Measure2: SUMX(Productie_OrganisatieGroei, CALCULATE(AVERAGE(Productie_OrganisatieGroei[AantalClienten]), Organisatie[Type] = "Woonlocatie"))

[Edit]

A few months later I participated in the 2-day SSAS Tabular workshop from Marco Russo and Alberto Ferrari. After the workshop I had a better understanding why measure 2 is working correctly, please see my follow-up post.