Can't use SELECTEDVALUE alternate value with Filter

Hi BradAF  - for reasons that someone smarter than I can explain, the context in which the SELECTEDVALUE measure is calculated changes when you create it as its own measure.  However, if you write your counting measure as this

MEASURE calculated_filtered[LastScanFindings] =
    CALCULATE (
        COUNT ( calculated_filtered[taskid] ),
        FILTER (
            calculated_filtered,
            calculated_filtered[scanned] = 
                  SELECTEDVALUE(calculated_filtered[scanned], MAX(calculated_filtered[scanned]))
        )
    )

or even this

MEASURE calculated_filtered[LastScanFindings] =
var __myDate = SELECTEDVALUE(calculated_filtered[scanned], MAX(calculated_filtered[scanned]))
return
    CALCULATE (
        COUNT ( calculated_filtered[taskid] ),
        FILTER (
            calculated_filtered,
            calculated_filtered[scanned] = __myDate 
        )
    )

You should get the results you are looking for, both with and without a date being selected.  DAX is funny that way in that splitting up the work into multiple measures does not always yield the right results.

Last date (RecCnt2 mirrors the code in this post)

Other Date

No Date

Hope this helps

David

dedelman_clng Is correct. When you call a measure inside an iterator, it performs a context transition that converts the row context inside FILTER to filter context for the measure. In each row inside the FILTER, there is only one date, so it just selects that value each time, which means the FILTER condition is true for each row.

Declaring it as a variable is a nice way of avoiding this problem since the variable is treated as a fixed value after it has been defined (i.e. it doesn't change depending on context later in the measure).

Hi BradAF  - every time CALCULATE or FILTER is called, the context gets swtiched. So your last CALCULATE(FILTER)) is called and evaluated for every row in your first FILTER. Not only would that make the calculation slower over larger data sets, but as you see you are getting incorrect results.

Best practice is to use variables to both keep the code more readable, and to keep from a value getting re-calculated inside an iterator of an iterator.  Also, by putting calculations that involve min/max into variables, you can use the built in filter functionality of CALCULATE and not have to write explicit FILTER commands.

LastWeekChange = 
    var __myDate = 
       SELECTEDVALUE(calculated_filtered[scanned], MAX(calculated_filtered[scanned]))
    var __LastWeek = 
        CALCULATE (
                  MAX(calculated_filtered[scanned]),
                  FILTER (                //You can also try removing this FILTER
                    calculated_filtered,  //and just having the < as a condition on CALCULATE
                    calculated_filtered[scanned] < __myDate
                         )
                   )
return
    calculated_filtered[LastScanFindings]
        - CALCULATE (
            COUNT(calculated_filtered[task]),
                calculated_filtered[scanned] = __LastWeek
          )

Hope this helps

David