Forum Discussion

than688's avatar
than688
Frequent Visitor
4 years ago

Matrix hiding records when using time intelligence, unless filtered. Table visual works fine.

Hello, wondering if anyone can help:

 

When using a matrix to compare Actuals with Prior Year via SAMEPERIODLASTYEAR, some categories of data are being hidden, but still included in totals.

 

I am using a marked date table created with CALENDARAUTO(3), and all relationships to my Data table are one>many, unidirectional filtering. The issue appears to occur when a category has no data in the selected year but does have data in the Prior Year. However:

  • Not all categories with the above criteria are missing - see 'new entities' in screenshot. There really are no rows of data here in Current year.
  • When I begin to filter the sub-categories, at some point of filtering the missing category reappears again.
  • Everything works fine in a table visual, so the issue seems specific to matrix.
  • My data only includes 1st of month dates, but my date table includes all dates.

My TI formula used in "Comparator":

CALCULATE(
[Act 000s], //<this is SUM ('Data All'[Actual])/1000
ALL ('Dates'[Date]),
SAMEPERIODLASTYEAR('Dates'[Date])
)
 
Screenshots:
PY shows correct totals but not everything they comprise:
Selecting the PY directly via the filter (which is a custom column in Date table) shows what was missing above:

If I go back to screenshot 1 and filter on subcategores/data in certain ways, I can get the missing categories to re-appear:

 

In contrast, the table visual seems to work fine:

 

I am unsure if this is a bug of some sort or whether I am missing some TI quirk, but I'm currently out of ideas. Thanks in advance for any help.

 

 

2 Replies

  • Anonymous's avatar
    Anonymous
    Not applicable

    llo

     

    Try to allow field with blank values on field I&E

  • than688's avatar
    than688
    Frequent Visitor

    EDIT: Amazingly, this has turned out to be an issue with the Matrix visual where the offending categories have been given a row height of one pixel - so they appear almost invisible (and are unviewable) but technically are present. The only way around this seems to be increasing the row header font size above 10, which corrects the behaviour for some reason.

     

    Has anyone encountered this before? Seems like it could be a bug.