Forum Discussion

Jkaelin's avatar
Jkaelin
Resolver I
9 years ago
Solved

Is this function written poorly?

Good afternoon,

 

I utilize the following DAX measure often. I am curious if this formula, from the naked eye, looks poorly written (poor performance) or if it looks fine given the information provided?

 

The data measure is crunching over 30,000 records.  If this question is not appropriate, or enough information given the question, please let me know and I will disregard.  Thank you.

 

12M T.R. Middle k ProductX :=
PRODUCTX (
    VALUES ( PrimaryDateID[PrimaryDateID] ),
    (
        1
            + (
                CALCULATE (
                    AVERAGE ( Performance[Month_3_Fwd_Blend_TR_then_PR_Ver_1#1] ),
                    FILTER (
                        Sheet1,
                        Sheet1[Total Return:Y-1]
                            >= PERCENTILE.INC ( Sheet1[Total Return:Y-1], MIN ( Quantile[Min_Value] ) )
                            && Sheet1[Total Return:Y-1]
                                <= PERCENTILE.INC ( Sheet1[Total Return:Y-1], MAX ( Quantile[Max_Value] ) )
                    )
                )
            )
                / 100
    )
)

Need Help

3 Replies

  • v-yulgu-msft's avatar
    v-yulgu-msft
    Microsoft Employee
    9 years ago

    Hi Jkaelin,

     

    What is your actually problem with above DAX formula? What did you mean "looks poorly written (poor performance)"?

     

    For its format, it looks tidy and clear. For its performance, it looks like not a complex logic. Acording to current information, this function look fine. Whether it is the best appropriate method, it depends on your detailed data in actual environment.

     

    Here are some articles written about optimizing DAX performance:

    Tune your PowerPivot DAX query, don’t use the entire table in a FILTER and replace SUMX if possible

    DAX Best Practices

    Optimize DAX expressions involving multiple measures

     

    Best regards,
    Yuliana Gu

    • Jkaelin's avatar
      Jkaelin
      Resolver I
      9 years ago

      v-yulgu-msft

       

      Thank you very much for the response.  I will read those articles.  I appreciate the feedback.  

       

      What did you mean "poor performance"?

      I use this measure in a Power Pivot table for quarterly periods from 2000-2017 (about 60 periods) and the time it takes to calculate is about 20 seconds.  If I use a derivative of that function to generate Stdev & Counts.  It takes a little bit longer.

       

      Thank you,

      James k

      • v-yulgu-msft's avatar
        v-yulgu-msft
        Microsoft Employee
        9 years ago

        Hi Jkaelin,

         

        Sometimes the long response time might also be related to the large amount of data.

         

        Regards,
        Yuliana Gu