ThetaSketches, retention analysis, and granularity

Given the retention analysis example from http://druid.io/docs/latest/development/extensions-core/datasketches-aggregators.html,
I’m trying to do something like this on my own. Now my problem is that I want to have an analysis pr day, and not the sum, e.g.

I want “granularity” to be “day”.

Problem seems to be, at least in my example, that the timestamps in “A_unique_users_week_1” are relative to the interval

[“2014-10-01T00:00:00.000Z/2014-10-07T00:00:00.000Z”],

whereas the timestamps in “A_unique_users_week_2” are relative to the interval

[“2014-10-08T00:00:00.000Z/2014-10-14T00:00:00.000Z”]

So, when using a granularity of “day” the post-aggregator is not able to find the intersection for the given day because the things in week_1

don’t have the same timestamps as the things in week_2.

I understand that this is reasonable, but I’d still like to have some sort of running/windowing retention rate.

Is this possible to do with druid?

Erik.