Aggr doesn't work with computed dimensions, so how do I do aggr(count(x), Date(timestamp))?

[#1]

Let's say, I need to find the avg number of distinct buyers per day so it should be, theoretically, something like this:

avg( aggr( count(distinct buyer_id), date(purchase_time) ) )

but this doesn't work because aggr() can't take computed dimension, I.E - date(purchase_timestamp).

What is the best solution for this?

Thanks in advance.

[mdmukramali · Accepted solution]

Dear,

can you try to create a field in the script level.

date(purchase_time) as Purchase_Date

then

avg( aggr( count(distinct buyer_id), Purchase_Date) )

hope it will work.

Thanks,

Mukram

[#3]

I was hoping to find a solution that does not involve without having to redefine the model but I guess there is none.

Thanks!