I want to analyse my user's behavior in my application against their "nativity" (If I started collecting data in January, a user with rank 0 is one who first started using the app in January, while a user with rank 4 is one who started this month).
I have 3 tables, relevant to this task:
1. events table; username column, action time column, calculated start of week column,..., calculated nativity rank column
2. users table; unique user names column, calculated # of rows of user in events table column, calculated nativity rank column (drawn from events table)
3. unique weeks table; unique start of week column
What I want is a graph whose x-axis is the nativity ranks (0-4), and whose values are the averages of the # of rows in events table of users with that rank.
Using all this calculated information I created a simple line graph with the x-axis the nativity ranks, and values of averages, as stated.
However, when I try to filter by weeks (*), the graph doesn't change much so I think the calculated column of # of rows doesn't change accoringly.
(*) I want to filter by weeks because by itself that graph isn't much of a surprise (almost declining throughout, which is expected)
I want to see how nativity effects their use in the last 2 weeks or so.
I have created the relations I think are appropriate for this architecture; which do you think will fix this?
Or does this form ofarchitecture not appropriate for this task?
So, any ideas why?
Thanks!
