Measure for counting unique users grouped by a parameter

[Anonymous]

Hi there!
Consider a table of facts. Each record contains user_id and some flag (true/false). I need to count unique users according to that flag and visualize it like a bar. For instance imagine that there're 10 users (10 records) with flag = "false" - (0, 10). After appending 2 records (12 records in total) where flag = "true" we get (2, 8), but not (2, 10). I came up with this measure which is as straight as possible, but gives no grouping at all.

unique_users = 
    VAR users_true = CALCULATETABLE(
        DISTINCT(users[user_id]),
        users[flag] = "true"
    )
    VAR users_false = CALCULATETABLE(
        DISTINCT(users[user_id])
    )

    RETURN SWITCH(
        TRUE(),
        "true" IN flags, COUNTROWS(users_true),
        COUNTROWS(users_false) - COUNTROWS(users_true)
    )

 

Thank you beforehand and best regards!

[v-yalanwu-msft · Accepted solution]

Hi, @Anonymous ;

Try it.

unique_users = 
VAR users_true =
    SUMMARIZE(FILTER('users', users[flag] = "true" ),[user_id])
VAR users_false =
    SUMMARIZE('users',[user_id])
RETURN
 COUNTROWS ( users_false ) - COUNTROWS ( users_true )

The final output is shown below:

Sorry I can't reproduce your data model, if this formula does not apply to your model, can you share simple files or scenarios?

Best Regards,
Community Support Team_ Yalan Wu
If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.