Running Total Distinct over time

Hi!

I have data for trainings/sessions that have been held over the last year, specifically who attended them.

I would like to visualize how many more unique attendees we are having every session, on each internal team. When I do a distinct count of attendees over time (a running total line graph), it gives me the running total sum of distinct attendees for each session, not session to session. The main question I would like to answer is: are the same people attending these sessions? Or are we attracting new people that are attending? If new people are attending, what team are they on?

Is what I am trying to visualize possible in Domo?

 

Here is an example use case:
Mary and Bill from Team A attended Session #1
Alex from Team B attended Session #1

Mary, Bill, and Jane attended Session #2

Alex from Team B attended Session #2

Team A had 2 unique attendees in Session #1 
Team B had 1 unique attendee in Session #1

 

Team A had 1 unique attendee in Session #2, Running Total: 3

Team B had 0 unique attendees in Session #2, Running Total: 1

Best Answer

  • user16684
    Accepted Answer

    I think I might use something like this in a MySql transform:

     

    SELECT *
    , (SELECT COUNT(1) FROM `attendees` p
    WHERE p.`Email`=s.`Email`
    AND p.`Event Date`<s.`Event Date`
    ) AS 'Previous Attendance Count'
    FROM `attendees` s

     

    This will give me a count of how many sessions they have attended, with the first one being '0'. Anything greater than 0 can be a non-unique attendee.

     

     

Answers

  • jaeW_at_Onyx
    jaeW_at_Onyx Budapest / Portland, OR 🟤

    I don't think any permutation of COUNT DISTINCT in beast modes will get you the answer you're looking for.

    I solve this problem with ETL like this.

    https://www.youtube.com/watch?v=Xb4QgKYgaqg

  • @jaeW_at_Onyx 

    I'm not sure if I totally understand how I could use the percent of the total. This would assume that I know how many people could possibly attend, which I do not.

     

    Basically, I need to know out of all the attendees for each session, how many have already attended any previous session.

This discussion has been closed.