Thanks <@U04B2K0S0PP> for the reply. I think that'...

# questionss

square-portugal-31785

05/15/2023, 10:25 AMThanks **@adventurous-megabyte-80699** for the reply. I think that's the current approach we're experimenting with, but I'm not sure how to do accurate averages, p99s, etc based on the aggregate properties.
For example, we want to determine the avg and p99 for publish time. Publish time is an aggregate property because it represents how long the entire publish took regardless of how many ranges were in the publish.
If every publish had the same number of ranges, I could use the built in avg and p99 function for the publish time property, but since the number of ranges varies, it skews the statistics.
• If publish 1 has 3 ranges and a separate event is logged for each range, we have 3 events with the same value for publish time. Let's say publish time is 1050.
• If publish 2 has 10 ranges, we have 10 ranges with the same publish time value. Let's say publish time is 200.
The average publish time should be 625, but Mixpanel reports 341.6 because the 10 200 values skew the numbers.
Any thoughts on how we could use the structure you suggested but still be able to aggregate analysis across all the publish events in Mixpanel?
Note: I figured out that I can do a distinct count based on a property, but I couldn't figure out how to do a distinct and an avg or p99 on the same value.