calculation

[MichielvS]

Spoiler

I want to calculate the average churn, based upon two tables in my model, containing:

Table1

Clientnr          Month         Subscription

A                     201501        1

A                     201502        1

B                     201501        1

B                     201502        1

Table2

Clientnr          Month         Churn

A                     201501        1

A                     201502        0

B                     201501        1

B                     201502        0

Resulting in a form of output where I can see that:

Month        Churn%

201501       100%

201502       0%

How do I do this in the report view?

[v-huizhn-msft · Accepted solution]

Hi @MichielvS,

Based on my understanding, for 201501 month, there is 2=1+1(AandB) Churn in Table2, 2=1+1(AandB) Subcription in table1, so 100%=2/2. And for 201502 month, there is 0=0+0(A and B) Churn in Table2, 2=1+1(A and B) Subcription in Table1, so 0%=0/2, right? If it is, please review the following steps.

First, click New table under Modeling in home page, create two new tables using the following formulas.

Table11 = SUMMARIZE(Table1,Table1[Month],"count",SUM(Table1[Subscription]))

Table22 = SUMMARIZE(Table2,Table2[Month],"count",SUM(Table2[Churn]))

Please see the new tables.

Then create relationship between Table11 and Table22.



In table22, create a calculated column using the formula, create a table visual to show the expected result.

Chrun% = Table22[count]/RELATED(Table11[count])

If this is not what you want, please let me know.

Best Regards,
Angelia

[MichielvS]

Now I would like to take this one step further, if possible. Do you know if I can incorporate dimensions to use as slicers, in one way or another? With the current result I can only show the totals over the base. What if for instance I would have the field 'gender' that I would like to be able to filter on in the output? Is that possible at all?

[v-huizhn-msft]

Hi @MichielvS,

Based on your description, it's difficult to reproduce the scenario. Unusualy there is one question in one case. I personally suggest you open a new thread and post the sample data and expexted result. Thanks for your understanding.

Best Regards,
Angelia

[MichielvS]

I will do that, thanks.

[MichielvS]

This is exactly what I was looking for! Thank you very much Angelia, your explanation is very helpful.