Hi @Vadym, are you using amazonbi account or other public preview account? We are rolling out a feature called Level-Aware-Calculation should be able to solve your problem. The functions will be available to all users soon around end of June. If you are able to use public preview, you can try the following functions: distinct_count({property_id},[{user_id}]), and drag that calculated field to the visual and select “average” as visual aggreagation. I re-produced your problem using a different dataset below, FYI (replacing industry by user, and property by order)
sorry, id[properties_PROD] is not the calculation formula, basically this is just a bad name in the real table. in my initial example this is just Property_Id, so just a unique ID per a property
We just launched the Level Aware Calculation feature for all users. Can you try the approach above again, using this function distinct_count({id[properties_PROD]},[{id}] )? I suspect the feature was not enabled for you at that time.
thanks @emilyzhu i will try this again and respond next week. meanwhile i have found a solution. for everyone who encounters this issue you can use this tutorial i drafted in Notion. also would be great if you can verify it @emilyzhu
@Vadym, That’s awesome! Your approach is correct, it uses the previous available functions- since distinctcountover creates duplication, you have to use rank and ifelse to de-dup the results. That is exactly why we are launching this new LAC feature. You can now directly use distinct_count(measure, [level]) to get the correct number at the [level] dimension, no need to de-dup anymore.
Please try the new approach and let us know how you think about it. Thanks!!