Looking to get some clarity around what the “Rollup” aggregate methodology is, and why it would differ from an Average aggregate function. See the attached Insight, where Average and Rollup are different, but very similar. I couldn’t find any specific information in the Documentation section.
Thank you in advance!
-Ma’ayan, Product Manager, GMV
Best answer by JakubView original
The rollup is an underlying calculation which acts as a smart aggregation and is calculated as if no attributes had been added to the report. It counts values that have missing references. Whereas the Sum counts what is actually being displayed in the report.
I will just add that same idea as Joe describes above between Sum and Rollup applies to Average as your original question was focused on the difference there. Basically average is simple function working with values in report and rollup works with underlying data and considers the metric. (e.g. you avoid counting average from averages) Allow me to add a link to our documentation with further clarification:
(the documentation I used is for our older UI so the screenshots do not match with your experience but the description is valid)
Hopefully it will answer your question but feel free to follow up with some example of yours if needed.
@Jakub and @Joseph Heun,
Thank you for the prompt response. I understand the differences between Rollup and Average now, I appreciate you linking the documentation to me. I will share that with the team once I get a little more clarity on the differences in the values between Rollup & Average.
A little background, we’re looking at transit agency data to assess total on-time performance (which stops drivers are leaving early, on-time, or late) across certain routes. Essentially, we’re counting <<how many stops were considered early / total number of stops>> to assess the performance over the course of a month to get these values.
So, the Avg aggregation is simply averaging the averages of each route (ie. 72.3% + 80% + … + 92.7% / # of observations) , while the Rollup function is averaging all the data points in each category. In this case, Rollup would give us a better indicator of overall service, since it would weight on-time performance more heavily for routes that operate more frequently and contribute more stops to the total # of stops.
Is my conclusion here correct? Let me know.
Hi Ma’ayn, your understanding seems correct here. If you are facing any issues with the calculation please let us know and we will be happy to directly address the function within the workspace.