Cohort analysis

[#1]

Hi friends,

trying to make a cohort table, where in the first column are cohorts (users by their registration date) and the other columns are # of users that made a target action, for example, item view.

I have the following data:

1. user registration date

2. user id

3. action date

4. action name

I created a staright tabe, where the dimension is Registration week (in script  WeekName(created_at/ 86400 + 25569) as Registration_week,) and expression is supposed to be number of unique users (registered in a certain week) that made at least one action in the first week after they registered on our site.

For exression I use =count(DISTINCT{$<action_date={">=$(min(action_date))<=$(date(min(action_date)+7))"}>}user_id)

And it returns number of unique users that registered in, for example, week 32 and that made actions but NOT in the first week after their registration, it counts it as of today.

Can't get where is my problem...

Table looks like this. In Cohort 30 only 208 unique users made actions, my table shows 287, i.e. overall number of unique users that made actions.

Some extracts from my script, maybe the problem is there:

Date(floor(timestamp/ 86400 + 25569)) as action_date,

WeekName(created_at/ 86400 + 25569) as Registration_week,

Thanks guys!