week on week % change beast mode

date examples are based on writing today - 6th September.

 

my aim is to create a beastmode, which can ultimately be applied to any metric, which shows the week-on-week % change. this value would be able to be displayed in a table by date(week) or as a KPI card which would be filtered to show data for last week.

the goal is to have the working beastmode for the values which are pre-calculated in the 'period over period' cards. I have attached the vairance bar line example I use - I would like the variance %s shown here as a beastmode value.

 

example:

a = sum the complete last week revenue (e.g. 26th August - 1st September  *Domo weeks run Sunday to Saturday)

b = sum the complete previous to last week revenue (e.g. 19th August - 25th August)

 

wow%change = sum(a) / sum(b) -1

 

e.g. 1500 / 1350 - 1 = 11.1%

 

 

I found some details on the domo knowledge board but cannot get the formula to work. I get a divide by zero error. example below...

 

(CASE
WHEN ((year(`activity_date`)=year(curdate())) AND (weekofyear(`activity_date`)=weekofyear(CURDATE())))
THEN `revenue`
END
/
CASE
WHEN (weekofyear(curdate())=1)
THEN (CASE WHEN ((year(`activity_date`)=(year(curdate()) - 1)) AND (weekofyear(`activity_date`)=52))
THEN `revenue` END )
ELSE (CASE WHEN ((year(`activity_date`)=year(curdate())) AND (weekofyear(`activity_date`)=(weekofyear(CURDATE()) -1)))
THEN `revenue` END )
END) -1

 

Comments

  • jstan
    jstan Contributor

    Hi @jonathanlooker, the reason you can get a divide by zero error is becuase you have division in your beast mode.  A denominator can never equal 0, else Domo throws an error.  

     

    My suggestion is you take the entire denominator and apply a case statement up front if the denominator = 0 then the value will be 0.  You numerators and denominators can contain additional case statements, you just need one that encompises the entire formula.

    CASE WHEN Denominator = 0 THEN 0 ELSE 

    Numerator

    /

    Denominator

    END

     

    Thanks,

    Josh

     

     

  • @jonathanlooker In addition to what @jstan said about adding a WHEN clause to prevent divide by zero errors, you have to make sure that the date range on your card is big enough to contain all data being compared. For example, if today (9/13/18) I am looking at data for last week Mon-Sun (9/3/18-9/9/18) vs the previous period (8/27/18-9/2/18), my card has to contain data from 8/27/18 and thereafter. Filtering the card for "last 22 days" or "last 3 weeks" should contain all of this data and then update automatically as you head into a new week.