Forum Discussion

devleena's avatar
devleena
Frequent Visitor
8 years ago
Solved

Incorrect Moving Average Date and Sum

Hello PowerBI Gurus,

 

I know moving average is a fairly common question but I have some current issues with my DAX  formula.

I will explain the data set I am using and the talk about the formula .

 

Say I have the following sample, the right most corner is the Rolling 7 day average.

 

So, basically if i have the following sets , the 3rd column ( 7 day rolling average) is populated from the 7th day observation and is the sum of the last 7 days .

The next value is the sum of the 2nd day to 7th day , and so on. The following is a screenshot of the formula worked out on excel.

 

 

I am using the following DAX formula for the Moving Average calculation, I need both SUM and Average, so the following 

 

 

Rolling 7 day SUM = SUMX(DATESINPERIOD('Date Key'[Date], LASTDATE('Date Key'[Date]),-7,DAY),[Total Sales])

 

Rolling 7 day MA = AVERAGEX(DATESINPERIOD('Date Key'[Date], LASTDATE('Date Key'[Date]),-7,DAY),[Total Sales])

 

The output of "Rolling 7 day SUM" is the following in the DAX for formula used above ( SUMX(DATESINPERIOD('Date Key'[Date], LASTDATE('Date Key'[Date]),-7,DAY),[Total Sales]))

 

 

 

You will see the first three rows for the 7 day match but then the next does not .

 

My table structure is ,

 

I have a main table named as  doolally_pos_master_staging

which has a date column named "pos_order_date" and a  sales column name "final_total"

 

and, TotalSales = SUM([final_total])

 

 

and,   Date Key is a Date table on the doolally_pos_master_staging table

 

Date Key = CALENDAR(min('DoolallySales doolally_pos_master_staging'[pos_order_date]), max('DoolallySales doolally_pos_master_staging'[pos_order_date]))

 

What am I doing wrong  and what should my formula be to ignore the first 7 days that does not have the rollup sum or rollup average.

 

Thanks you very much for your help.

 

Regards,

Dev

Need Help
Show and Tell
Tips and Tricks
  • v-yulgu-msft's avatar
    v-yulgu-msft
    8 years ago

    Hi devleena,

     

    Please try this solution to calculate moving average without adding a calendar table.

     

    In Query Editor mode, sort the field 'DoolallySales doolally_pos_master_staging'[pos_order_date] in Ascending order. Then, add an index column.

     

    Then, in report view, add DoolallySales doolally_pos_master_staging'[pos_order_date] into table visual, create measues like this:

    TotalSales = SUM([final_total])

Rolling 7 day SUM2 =
IF (
    MAX ( doolally_pos_master_staging[Index] ) < 7,
    BLANK (),
    CALCULATE (
        [TotalSales],
        FILTER (
            ALL ( doolally_pos_master_staging ),
            doolally_pos_master_staging[Index] <= MAX ( doolally_pos_master_staging[Index] )
                && doolally_pos_master_staging[Index]
                    > MAX ( doolally_pos_master_staging[Index] ) - 7
        )
    )
)

Rolling 7 day Average2 =
IF (
    MAX ( doolally_pos_master_staging[Index] ) < 7,
    BLANK (),
    CALCULATE (
        [TotalSales],
        FILTER (
            ALL ( doolally_pos_master_staging ),
            doolally_pos_master_staging[Index] <= MAX ( doolally_pos_master_staging[Index] )
                && doolally_pos_master_staging[Index]
                    > MAX ( doolally_pos_master_staging[Index] ) - 7
        )
    )
)
    / 7

     

    Best regards,

    Yuliana Gu

5 Replies

    • devleena's avatar
      devleena
      Frequent Visitor
      8 years ago

      Hello Greg,

       

      Thank you for the prompt reply .

       

      Please find the text data below 

       

      DateTotal Sales7 Day Rolling Avg
      13/11/15 0:00₹31,650 
      14/11/15 0:00₹1,07,740 
      15/11/15 0:00₹1,38,140 
      16/11/15 0:00₹52,370 
      17/11/15 0:00₹71,570 
      18/11/15 0:00₹1,25,490 
      19/11/15 0:00₹88,890₹6,15,850
      20/11/15 0:00₹1,50,490₹7,34,690
      21/11/15 0:00₹1,50,520₹7,77,470
      23/11/15 0:00₹83,270₹7,22,600
      24/11/15 0:00₹1,15,200₹7,85,430
      25/11/15 0:00₹89,170₹8,03,030
      26/11/15 0:00₹76,920₹7,54,460
      27/11/15 0:00₹2,12,510₹8,78,080
      28/11/15 0:00₹1,51,440₹8,79,030
      29/11/15 0:00₹1,46,350₹8,74,860
      30/11/15 0:00₹77,640₹8,69,230
      01/12/15 0:00₹65,970₹8,20,000
      02/12/15 0:00₹1,24,530₹8,55,360
      03/12/15 0:00₹87,100₹8,65,540
      04/12/15 0:00₹1,57,820₹8,10,850
      05/12/15 0:00₹2,05,070₹8,64,480
      06/12/15 0:00₹1,19,550₹8,37,680
      07/12/15 0:00₹87,170₹8,47,210
      08/12/15 0:00₹93,570₹8,74,810
      09/12/15 0:00₹1,50,260₹9,00,540
      10/12/15 0:00₹1,49,450₹9,62,890
      11/12/15 0:00₹2,36,620₹10,41,690
      12/12/15 0:00₹1,86,540₹10,23,160
      13/12/15 0:00₹1,37,470₹10,41,080

       

      Hope this helps 

       

      Thanks and Regards,

      Dev

    • devleena's avatar
      devleena
      Frequent Visitor
      8 years ago

      The amount being calculated is correct . There is a gap in the record , 22nd Novemeber 2015 is not there.

      So I switched on the key for showing null values and the math is now working fine.

       

      What I still not being able to do is, ignore the first 7 days where there is no rollup sum. 

    • devleena's avatar
      devleena
      Frequent Visitor
      8 years ago

      Hello Greg,

       

      I tried your solution and a little RnD at my end too.

       

      However, I am not being able to calculate the Moving Average as they get calculated in Microsoft Excel .

       

      The Moving Average is being calculated correctly now, however I do not want to put the first 7 non-average days in the graph .

      I can drop them ( exclude rows ) manually,  however, it will not really solve my case , because the start date ( earliest date ) of my dataset varies .

       

      So, say when I am calculating the 7  day moving average for a store in Location A, the earliest day is Nov 13th , 2016 .

      The average days start from Nov 19th ( 7 days from start ) and I am expecting my result to output only days from Nov 19th and onwards.

       

      I also have a different store in a different location, the earliest day of which is  Jan 22nd, 2017, so the 7th day will be different.

       

       

      Please help on what is the best solution to this.

       

       

      Thanks and Regards,

      Dev

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

        Hi devleena,

         

        Please try this solution to calculate moving average without adding a calendar table.

         

        In Query Editor mode, sort the field 'DoolallySales doolally_pos_master_staging'[pos_order_date] in Ascending order. Then, add an index column.

         

        Then, in report view, add DoolallySales doolally_pos_master_staging'[pos_order_date] into table visual, create measues like this:

        TotalSales = SUM([final_total])

Rolling 7 day SUM2 =
IF (
    MAX ( doolally_pos_master_staging[Index] ) < 7,
    BLANK (),
    CALCULATE (
        [TotalSales],
        FILTER (
            ALL ( doolally_pos_master_staging ),
            doolally_pos_master_staging[Index] <= MAX ( doolally_pos_master_staging[Index] )
                && doolally_pos_master_staging[Index]
                    > MAX ( doolally_pos_master_staging[Index] ) - 7
        )
    )
)

Rolling 7 day Average2 =
IF (
    MAX ( doolally_pos_master_staging[Index] ) < 7,
    BLANK (),
    CALCULATE (
        [TotalSales],
        FILTER (
            ALL ( doolally_pos_master_staging ),
            doolally_pos_master_staging[Index] <= MAX ( doolally_pos_master_staging[Index] )
                && doolally_pos_master_staging[Index]
                    > MAX ( doolally_pos_master_staging[Index] ) - 7
        )
    )
)
    / 7

         

        Best regards,

        Yuliana Gu