Forum Discussion

AnthNC's avatar
AnthNC
Helper II
3 years ago
Solved

Variance calculation in Table or Matrix

Hi everyone

 

I'm doing a sales comparison report in which the end user can pick any 2 dates in a slicer (2016 and 2021 in below example).

To show the results, I'm trying to achieve this visual made with excel :

 

In PBI I used a table and a matrix to visualize the results but i can't replicate the excel table above.

 

1/ TABLE

I went first for a table for which I created 3 measures :

a) SalesTotal_YearOldest = CALCULATE([SalesTotal], FILTER(CALENDRIER, CALENDRIER[Year]=MIN(CALENDRIER[Year] ))) ----- to show 2016 results

b) SalesTotal_YearMostRecent = CALCULATE([SalesTotal], FILTER(CALENDRIER, CALENDRIER[Year]=MAX(CALENDRIER[Year] ))) ----- to show 2021 results

c) Variance = [SalesTotal_YearMostRecent] - [SalesTotal_YearOldest] / [SalesTotal_YearOldest]

 

Here's the visual :

This table is ok but i can't find a way to have dynamic column headers : YEAR(oldest) would be 2016 and YEAR(most recent) would be 2021 (based on the years picked in the slicer).

I've read many posts on dynamic column headers but it seems it's not possible for tables (unless you have an idea on how to do it ?).

 

So i opted for a matrix.

 

2/ MATRIX

Here's the visual without the Variance measure :

Problem is i can't use the Variance measure in that matrix. The measure won't stay in the "Colonnes" field when i try to drag and drop the measure in. And when i put it in the "Valeurs" field, here's what it does :

 

 

Can anyone help me out reproducing the screenshot excel table please ?

Here's the data model if it can be of any help :

 

I hope my explanations are clear and detailed enough.

Thank you 🙏

 

  • v-jianboli-msft's avatar
    v-jianboli-msft
    3 years ago

    Hi AnthNC ,

     

    Based on your description, I have created a simple example:

    Please try:

    Measure = 
var _a = CALCULATE(SUM('Table'[Value]),FILTER('Table',[Year]=MIN('Table'[Year])))
var _b = CALCULATE(SUM('Table'[Value]),FILTER('Table',[Year]=MAX('Table'[Year])))
return 
IF(ISINSCOPE('Table'[Year]),SUM('Table'[Value]),FORMAT(DIVIDE(_b-_a,_a),"Percent"))

    Change the name of  column subtotal:

    Final output:

    Best Regards,

    Jianbo Li

    If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

4 Replies

  • v-jianboli-msft's avatar
    v-jianboli-msft
    Community Support
    3 years ago

    Hi AnthNC ,

     

    Based on your description, I have created a simple example:

    Please try:

    Measure = 
var _a = CALCULATE(SUM('Table'[Value]),FILTER('Table',[Year]=MIN('Table'[Year])))
var _b = CALCULATE(SUM('Table'[Value]),FILTER('Table',[Year]=MAX('Table'[Year])))
return 
IF(ISINSCOPE('Table'[Year]),SUM('Table'[Value]),FORMAT(DIVIDE(_b-_a,_a),"Percent"))

    Change the name of  column subtotal:

    Final output:

    Best Regards,

    Jianbo Li

    If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

    • AnthNC's avatar
      AnthNC
      Helper II
      3 years ago

      Many thanks Jianbo Li !

      I'm finding out about how to use variables in measures. Thanks again

    • eddiedee's avatar
      eddiedee
      Regular Visitor
      2 years ago

      This is a very clear example thanks for that.

       

      Question, I'm computing a numeric rather than percentage variance. In my (very large) dataset there may be instances where the "subcategory" isn't present in a "year". How can I incorporate COALESCE into the above in order to compute a variance where for a given "subcategory", one "year" is BLANK?

       

      I've tried inserting in both the variables and in the IF(INSCOPE line without any success.

       

      subcategory20162018Variance
      a 9090
      b208060
      c307040

       

      I would've thought this usage of COALESCE would accomplish my objective but I'm stuck:

       

      Measure =
      var _a = COALESCE(CALCULATE(SUM('Table'[Value]),FILTER('Table',[Year]=MIN('Table'[Year]))),0)
      var _b = COALESCE(CALCULATE(SUM('Table'[Value]),FILTER('Table',[Year]=MAX('Table'[Year]))),0)
      return
      IF(ISINSCOPE('Table'[Year]),SUM('Table'[Value]),COALESCE(_b,0) - COALESCE(_a,0))
    • PBIdashboards's avatar
      PBIdashboards
      Post Patron
      2 months ago

      The dynamic column headers problem you hit is a known limitation native Table visual can't do it, and Matrix requires fixed column definitions. The ISINSCOPE workaround in the accepted solution is the cleanest DAX approach.

      For the end-user date picker + dynamic variance combination you're describing, this is exactly the use case Flexa Tables (AppSource) was built for users select any two periods to compare, variance columns appear automatically with dynamic headers. No DAX needed for the display layer. Search "Flexa Tables" on Microsoft AppSource.