Rolling 12 Month Avg

Hi all - Currently I have a dataset that contains a date (by quarters), brand, sub-segment, sample type, and a NPS column. Using a Beastmode calculation, I take the favorable responses and divide it by the total number of responses to get a NPS score. This data is granular (many NPS respondents, by brand, by segment per quarter).

I load this dataset in Domo, apply necessary filters, and I am able to produce a graph using the Beastmode calculation that shows the NPS score by quarter. Using this Beastmode calculation, I am able to slice and dice the data.


NPS score Beastmode calculation:

count(case when `T2B` = 'Yes' then 1 end) / count(case when `Month_Year` is not null then 1 end)


The problem I'm trying to tackle is that the NPS score should be a rolling 12 month average (past 4 quarters).

Ideally, it would be nice to resolve my issue using Beastmode, but I'm open to any other options! I can foresee if an ETL is used, some of this flexibility would be eliminated as things would now be hardcoded.


Thanks in advance and pls let me know if you need anymore info :)



Best Answers

  • GrantSmith
    GrantSmith Coach
    Answer ✓

    Hi @DomoNewbie

    Have you read my prior article on rolling averages using window functions? This may help you out: https://dojo.domo.com/discussion/52679/domo-ideas-conference-beast-modes-rolling-averages#latest

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

    How far back does your data go? If you don't have any data prior to 2020-Q4 it'll return NULL because there are no prior rows for it to pull from. When you have a NULL with a simple arithmetic operator (=, -, *, /) the entire result ends up being NULL. You can wrap your entire LAG...OVER() in a COALESCE function and coalesce it with 0 to force NULLs to be 0

    COALESCE(lag(count(... over (order by `Quarter`), 0)
    
    **Was this post helpful? Click Agree or Like below**
    **Did this solve your problem? Accept it as a solution!**

Answers

  • GrantSmith
    GrantSmith Coach
    Answer ✓

    Hi @DomoNewbie

    Have you read my prior article on rolling averages using window functions? This may help you out: https://dojo.domo.com/discussion/52679/domo-ideas-conference-beast-modes-rolling-averages#latest

    **Was this post helpful? Click Agree or Like below**
    **Did this solve your problem? Accept it as a solution!**
  • Hi Grant - I read your post, modified the formula found in the linked post to suit my needs:

    Which yielded the below results. Awesome. Exactly what I needed! Thanks for your guidance and quick response!


    Do you happen to know why the quarters with less than 4 scores do not have any rolling avgs? Thanks again!

  • GrantSmith
    GrantSmith Coach
    Answer ✓

    How far back does your data go? If you don't have any data prior to 2020-Q4 it'll return NULL because there are no prior rows for it to pull from. When you have a NULL with a simple arithmetic operator (=, -, *, /) the entire result ends up being NULL. You can wrap your entire LAG...OVER() in a COALESCE function and coalesce it with 0 to force NULLs to be 0

    COALESCE(lag(count(... over (order by `Quarter`), 0)
    
    **Was this post helpful? Click Agree or Like below**
    **Did this solve your problem? Accept it as a solution!**
  • Perfect. Your response makes sense. I only loaded 5 quarters worth of data into Domo which explains why the formula returns NULL. I was thinking that the remaining quarters that had data still would be able to divide by 4, but the previous month's NULL errored out the result.


    Thanks again for your expertise and responsiveness 😀