Skip to main content
cancel
Showing results for 
Search instead for 
Did you mean: 

Earn the coveted Fabric Analytics Engineer certification. 100% off your exam for a limited time only!

Reply
Anonymous
Not applicable

Measure Optimization - IF() Reduction

Hello - I am trying to see if I can reduce the number of steps in the following measure.  Sub Measures are as follows:

 

[EstimatedTotalCost] is the sum of 4 individual columns that sit in the same table.
[SumLaborCost], [SUMSubCosts], [SumMaterialCost] are each a sum of a single columns.

 

In summary, the measure goes through a few logic steps to determine if an estimate is a "Complete Estimate", an estimate with "Material, no Labor", or an estimate that has "No Material, No Labor".  These are determined by the correlating columns having a value >0, or =0. 

The measure works as needed, but wondering if there's a more efficient way to return the desired outcome.  I am not able to add any columns to the data set, so I have to get where I'm going with a measure.

EstimateQuality = 
var estimatedtotalcost = [EstimatedTotalCost] 
var estimatedlaborcost = [SUMLaborCost] 
var estimatedsubcost = [SUMSubCosts] 
var estimatedmaterialcost = [SUMMaterialCost]
return  
    IF(
        ISBLANK(estimatedtotalcost),
        BLANK(),
            IF(estimatedsubcost > 0,
            "Complete Estimate",
            IF(estimatedmaterialcost>0
                &&estimatedlaborcost=0,
            "Material, No Labor",
            IF(estimatedtotalcost=0,"No Material, No Labor","Complete Estimate")
            )
          )
        )

 @Vera_33 

1 ACCEPTED SOLUTION

Hi @Anonymous 

 

So you go with one measure, the same test

Vera_33_0-1622250486891.png

test = 
VAR T1 =
    GENERATE (
        GROUPBY ( SalesHistory, SalesHistory[SO] ),
        VAR EstimatedLaborCost =
            CALCULATE ( SUM ( SalesHistory[Estimated LaborCost] ) )
        VAR EstimatedSubCost =
            CALCULATE ( SUM ( SalesHistory[Estimated SubcontractorCost] ) )
        VAR EstimatedMaterialCost =
            CALCULATE ( SUM ( SalesHistory[Estimated MaterialCost] ) )
        VAR EstimatedOtherCost =
            CALCULATE ( SUM ( SalesHistory[Estimated OtherCost] ) )
        VAR EstimatedTotalCost = EstimatedLaborCost + EstimatedSubCost + EstimatedMaterialCost + EstimatedOtherCost
        VAR EstimateQuality1 =
            SWITCH (
                TRUE (),
                ISBLANK ( EstimatedTotalCost ), BLANK (),
                ( EstimatedTotalCost > 0
                    && EstimatedMaterialCost = EstimatedTotalCost )
                    || EstimatedTotalCost = 0, "Incomplete",
                "Complete"
            )
        RETURN
            ROW ( "EstimateQuality1", EstimateQuality1 )
    )

VAR CompleteCount =    COUNTROWS ( FILTER ( T1, [EstimateQuality1] = "Complete" ) )
VAR TotalSO = COUNTROWS(VALUES(SalesHistory[SO]))
RETURN
DIVIDE(CompleteCount,TotalSO)

 

View solution in original post

11 REPLIES 11
v-stephen-msft
Community Support
Community Support

Hi @Anonymous ,


Could you tell me if your problem has been solved?
If it is, kindly Accept it as the solution. More people will benefit from it.
Or you are still confused about it, please provide me with more details about your problem.


Best Regards,
Stephen Tao

Anonymous
Not applicable

Since posting, I've gone another direction, using a combination of SWITCH() and combining all measures into multiple VARs.

 

To summarize, the measure is calculating total costs (EstimatedTotalCost), determining if an estimate is "Complete" or Incomplete" (Estimate Quality), Counting all estimates (CountAllEstimates), counting all complete estimates (CompleteEstimateCount), and calculating the ratio of complete to total estimates (EstimateQuality%).

 

In silos, each of the VARs are working as expected, except for CompleteEstimateCount. It counts all rows, not just the rows that EstaimteQuality is calling "Complete".

 

When I drop the EstimateQuality measure into a data table, it does call out the Sales Orders (SOs) that are and are not "Complete".

 

This leads me to believe that there is an issue with my COUNTAX() measure.  I have to use Distinctcount() for 'Sales History'[SO] -> There are cases where the SO has multiple service lines.  In those cases, there are two of the same SOs, but I don't want to count it as two.  Using Distinctcount() fixes that problem.

 

In summary, all of the VARs work in their own silos, when I try to count the "Complete" SOs, the measure counts all of the SOs, not just the ones that EstimateQuality has determined as complete.

 

@Greg_Deckler 

Kudo's to the following post for getting me down this thought process. 
https://community.powerbi.com/t5/Community-Blog/Performance-Tuning-DAX-Part-1/ba-p/976275

 

 

 

Estimatequality% = 

VAR EstimatedLaborCost =
    SUM ( SalesHistory[Estimated LaborCost] )

VAR EstimatedSubCost =
    SUM ( SalesHistory[Estimated SubcontractorCost] )

VAR EstimatedMaterialCost =
    SUM ( SalesHistory[Estimated MaterialCost] )

Var EstimatedOtherCost = 
    SUM(Sales[Estimated OtherCost])

Var EstimatedTotalCost =
SUMX (
        SalesHistory,
        EstimatedLaborCost + EstimatedSubCost + EstimatedMaterialCost + EstimatedOtherCost
    )

Var EstimateQuality =
SWITCH(
    TRUE(),
        ISBLANK(EstimatedTotalCost),BLANK(),
        EstimatedTotalCost > 0
            && EstimatedMaterialCost = EstimatedTotalCost
            || EstimatedTotalCost = 0 , "Incomplete",
            "Complete"
)

Var CountAllEstimates = 
DISTINCTCOUNT(SalesHistory[SO])

Var CompleteEstimateCount = 
    COUNTAX(FILTER(SalesHistory,estimatequality="Complete"),DISTINCTCOUNT(SalesHistory[SO]))
    
Return

DIVIDE(CompleteEstimateCount,CountAllEstimates)

 

@Vera_33 

Hi @Anonymous 

 

So you still have problem with this measure? Did not get what you want?

Can you provide some sample data with expected result? It is difficult to understand just based on a measure...

Var EstimateQuality =
SWITCH(
    TRUE(),
        ISBLANK(EstimatedTotalCost),BLANK(),
        (EstimatedTotalCost > 0
            && EstimatedMaterialCost = EstimatedTotalCost)
            || EstimatedTotalCost = 0 , "Incomplete",
            "Complete"
)

Var CountAllEstimates = 
COUNTROWS(VALUES(SalesHistory[SO]))

 

Anonymous
Not applicable

@Vera_33 Of course.  I've uploaded it to GDocs.

 

Sample Data: https://drive.google.com/file/d/1Uu53QZFkrziBEdoLzORoY__dWnFJlnsj/view?usp=sharing

 

Sample PBI File: https://drive.google.com/file/d/1yxISUan6KIdk6nVKMUuKQTSGSLo80IYF/view?usp=sharing

 

The table is returning what I expect in the EstimateQualityCheck measure (complete/incomplete).  However, the aggregation isn't correct in the CompleteCount measure.

 

I'd expect that the 8 "Incompletes" would be excluded from the CompleteCount measure.  

 

The final calculation should return something in the ballpark of (147-8)/147 = 94.5%

 

In other words, why aren't these incomplete lines removing themselves from the calculation?

 

I also can't figure out why the CompleteCount is 148, not 147.  The Distinctcount([SalesHistory]SO) is working in the TotalCount measure.  But not when it's nested in the CompleteCount measure.

 

Fischelsr01_0-1622036032084.png

 

Thanks in advance for your continued support.

Hi @Anonymous 

 

I am running into another direction.

 

Vera_33_0-1622085927347.png

If you are not calling a measure, but write them, wrap CALCULATE otherwise you sum up the whole table

test =
VAR T1 =
    GENERATE (
        GROUPBY ( SalesHistory, SalesHistory[SO] ),
        VAR EstimatedLaborCost =
            CALCULATE ( SUM ( SalesHistory[Estimated LaborCost] ) )
        VAR EstimatedSubCost =
            CALCULATE ( SUM ( SalesHistory[Estimated SubcontractorCost] ) )
        VAR EstimatedMaterialCost =
            CALCULATE ( SUM ( SalesHistory[Estimated MaterialCost] ) )
        VAR EstimatedOtherCost =
            CALCULATE ( SUM ( SalesHistory[Estimated OtherCost] ) )
        VAR EstimatedTotalCost = EstimatedLaborCost + EstimatedSubCost + EstimatedMaterialCost + EstimatedOtherCost
        VAR EstimateQuality1 =
            SWITCH (
                TRUE (),
                ISBLANK ( EstimatedTotalCost ), BLANK (),
                ( EstimatedTotalCost > 0
                    && EstimatedMaterialCost = EstimatedTotalCost )
                    || EstimatedTotalCost = 0, "Incomplete",
                "Complete"
            )
        RETURN
            ROW ( "EstimateQuality1", EstimateQuality1 )
    )
RETURN
    COUNTROWS ( FILTER ( T1, [EstimateQuality1] = "Complete" ) )

 

 If you do have those measures, you can simply call it

test1 =
VAR T1 =
    ADDCOLUMNS ( VALUES ( SalesHistory[SO] ), "test111", [EstimateQualityCheck] )
RETURN
    COUNTROWS ( FILTER ( T1, [test111] = "Complete" ) )

I am not sure how your measure was working - COUNTAX part, but it basically counts the total row of your dummy data table -  there is one SO with 2 entries

Anonymous
Not applicable

That sure does work!  Thanks so much.

 

One last thing.  How would you incorporate a final output of:

Divide([Test],[TotalCount]... But do it in the same measure.

Hi @Anonymous 

 

So you go with one measure, the same test

Vera_33_0-1622250486891.png

test = 
VAR T1 =
    GENERATE (
        GROUPBY ( SalesHistory, SalesHistory[SO] ),
        VAR EstimatedLaborCost =
            CALCULATE ( SUM ( SalesHistory[Estimated LaborCost] ) )
        VAR EstimatedSubCost =
            CALCULATE ( SUM ( SalesHistory[Estimated SubcontractorCost] ) )
        VAR EstimatedMaterialCost =
            CALCULATE ( SUM ( SalesHistory[Estimated MaterialCost] ) )
        VAR EstimatedOtherCost =
            CALCULATE ( SUM ( SalesHistory[Estimated OtherCost] ) )
        VAR EstimatedTotalCost = EstimatedLaborCost + EstimatedSubCost + EstimatedMaterialCost + EstimatedOtherCost
        VAR EstimateQuality1 =
            SWITCH (
                TRUE (),
                ISBLANK ( EstimatedTotalCost ), BLANK (),
                ( EstimatedTotalCost > 0
                    && EstimatedMaterialCost = EstimatedTotalCost )
                    || EstimatedTotalCost = 0, "Incomplete",
                "Complete"
            )
        RETURN
            ROW ( "EstimateQuality1", EstimateQuality1 )
    )

VAR CompleteCount =    COUNTROWS ( FILTER ( T1, [EstimateQuality1] = "Complete" ) )
VAR TotalSO = COUNTROWS(VALUES(SalesHistory[SO]))
RETURN
DIVIDE(CompleteCount,TotalSO)

 

Anonymous
Not applicable

@Vera_33 That's perfect.

 

When incorporating with the rest of my company data, I went from ~33 seconds per operation to 8.9.  Fantastic!

Vera_33
Resident Rockstar
Resident Rockstar

Hi @Anonymous 

 

It does not matter whether the code is long or short...but the performance and the side effect... so have you tried using SWITCH instead of nested IFs? Simply change it, not sure if it is working for your logic...have a test

 

EstimateQuality =
VAR estimatedtotalcost = [EstimatedTotalCost]
VAR estimatedlaborcost = [SUMLaborCost]
VAR estimatedsubcost = [SUMSubCosts]
VAR estimatedmaterialcost = [SUMMaterialCost]
RETURN
    SWITCH (
        TRUE (),
        ISBLANK ( estimatedtotalcost ), BLANK (),
        estimatedsubcost > 0, "Complete Estimate",
        estimatedmaterialcost > 0
            && estimatedlaborcost = 0, "Material, No Labor",
        estimatedtotalcost = 0, "No Material, No Labor",
        "Complete Estimate"
    )

 

@Johanno good point to always use DAX formatter😁

Johanno
Responsive Resident
Responsive Resident

If it works I think you should be satisfied and move on to the next challange. 😀 There's nothing wrong using nested IF, sometimes you can avoid them by using SWITCH but SWITCH doesn't make any difference other than understanding the code since it's internally translated into nested IFs.

 

Sure, if you could add a column in Power Query or in the data source you might write it differently. The only thing I note is the formatting of the code (DAX Formatter is a great tool!) so it's easier to see which IFs are subordinate:

 

 

EstimateQuality =
VAR estimatedtotalcost = [EstimatedTotalCost]
VAR estimatedlaborcost = [SUMLaborCost]
VAR estimatedsubcost = [SUMSubCosts]
VAR estimatedmaterialcost = [SUMMaterialCost]
RETURN
    IF (
        ISBLANK ( estimatedtotalcost ),
        BLANK (),
        IF (
            estimatedsubcost > 0,
            "Complete Estimate",
            IF (
                estimatedmaterialcost > 0
                    && estimatedlaborcost = 0,
                "Material, No Labor",
                IF ( estimatedtotalcost = 0, "No Material, No Labor", "Complete Estimate" )
            )
        )
    )

 

 

Anonymous
Not applicable

Thanks, @Johanno   Definetly using Dax Formatter to accomplish this nicely cascaded format 🙂

Helpful resources

Announcements
April AMA free

Microsoft Fabric AMA Livestream

Join us Tuesday, April 09, 9:00 – 10:00 AM PST for a live, expert-led Q&A session on all things Microsoft Fabric!

March Fabric Community Update

Fabric Community Update - March 2024

Find out what's new and trending in the Fabric Community.