Beast Mode to show total for previous week and two weeks prior

I am trying to create a table that has a column that shows a total from the previous week and a column with a total from 2 weeks prior. The data crosses the new year, 2020 to 2021. When I create the calculation for 2 weeks prior, it yields an accurate result but when I try a similar calculation for the previous week, it is not accurate. 

 

Here is the calculation I am using for the sum of visit 2 weeks prior:

Sum(CASE WHEN WEEK(DATE_SUB(Current_Date(), INTERVAL 2 WEEK)) = WEEK(`Visit Start Date`) AND YEAR(DATE_SUB(Current_Date(), INTERVAL 2 WEEK)) = YEAR(`Visit Start Date`) THEN 1 Else 0 End)

 

I tried to get the total for the previous week using the calculation below and it is not yielding the correct results. 

Sum(CASE WHEN WEEK(DATE_SUB(Current_Date(), INTERVAL 1 WEEK)) = WEEK(`Visit Start Date`) AND YEAR(DATE_SUB(Current_Date(), INTERVAL 1 WEEK)) = YEAR(`Visit Start Date`) THEN 1 Else 0 End)

 

What calculations would be best for this scenario?

Comments

  • You could try setting up your comparison using the YEARWEEK function instead of WEEK and YEAR separately. So your beast mode would look like this:

     

    Sum(CASE WHEN YEARWEEK(DATE_SUB(Current_Date(), INTERVAL 2 WEEK)) = YEARWEEK(`Visit Start Date`) THEN 1 Else 0 End)

  • Hi @user028686 

     

    Whenever I typically need to do some period over period comparisons I actually restructure my data so that I have a report date and a "comparison date" (the actual date the metric was recorded on) and calculate it as necessary. This allows you to have custom defined periods (like 2 weeks ago and a week ago).

     

    You'd essentially use the report date in your graph but then have beast modes to determine which type of period to display (last week, 2 weeks ago etc.) and centralizes this logic so you have one place to maintain it rather than across multiple cards and beast modes.

     

    I've written about this method previously here: https://dojo.domo.com/t5/Card-Building/show-percent-on-period-over-period-graph/m-p/50540/highlight/true#M7313

    **Was this post helpful? Click Agree or Like below**
    **Did this solve your problem? Accept it as a solution!**
  • Thank you for your response. This worked to show 2 weeks prior but when I do that to only show the previous week, it just displays the total for 2020 data, not 2021. The previous week is 12/27/20 - 1/2/21. 

  • @user028686 

     

    If you absolutely must use a beast mode you could utilize the DAYOFWEEK function to get the end of the prior week and then subtract the number of weeks you're interested in.  DAYOFWEEK returns values between 1 (Sunday) and 7 (Saturday). Subtracting that value for the number of days will return the end of last week

     

    For example - 1 Week Ago / Last Week:

     

    SUM(
    CASE WHEN (`Date`) > DATE_SUB(CURRENT_DATE(), INTERVAL DAYOFWEEK(CURRENT_DATE()) + (1*7) DAY)
    AND (`Date`) <= DATE_SUB(CURRENT_DATE(), INTERVAL DAYOFWEEK(CURRENT_DATE()) + (0*7) DAY)
    THEN 1
    ELSE 0
    END)

     

     

    2 Weeks Ago:

     

    SUM(
    CASE WHEN (`Date`) > DATE_SUB(CURRENT_DATE(), INTERVAL DAYOFWEEK(CURRENT_DATE()) + (2*7) DAY)
    AND (`Date`) <= DATE_SUB(CURRENT_DATE(), INTERVAL DAYOFWEEK(CURRENT_DATE()) + (1*7) DAY)
    THEN 1
    ELSE 0
    END)

     

     

    To break it down for the week ago example: DATE_SUB(INTERVAL DAYOFWEEK(CURRENT_DATE()) + (0*7) DAY) says give me the Date of the Saturday before today. (0*7 here isn't needed but is to help illustrate the logic saying I don't want any weeks prior to this date)

     

    (1*7) adds an extra week before that. (2*7) is two weeks prior.

    **Was this post helpful? Click Agree or Like below**
    **Did this solve your problem? Accept it as a solution!**
  • Again, I'd highly recommend you configure an offset dimension to setup a week prior and 2 weeks prior custom offsets and structure your data as such.

     

    @jaeW_at_Onyx did a great video explaining this whole process as well: https://www.youtube.com/watch?v=CDKNOmKClms 

    **Was this post helpful? Click Agree or Like below**
    **Did this solve your problem? Accept it as a solution!**