Kite state transition from "Resurrected" to "New"
Many users are moving from "Resurrected" to "New".
Example of users with this pattern in August 2019:
+-------------------+-------------------------------+
|user_guid_kite |journey_string |
+-------------------+-------------------------------+
|1003452005647130643|resurrected, new, curious, cold|
|1003452092720881668|resurrected, new, curious, cold|
|1003452215974699011|resurrected, new, curious, cold|
|1003452303409160198|resurrected, new, curious, cold|
|1003452584448499723|resurrected, new, curious, cold|
|1003452662076678154|resurrected, new, curious, cold|
|1003452714325123076|resurrected, new, curious, cold|
|1003452818838790156|resurrected, new, curious, cold|
|1003452841710329863|resurrected, new, curious, cold|
|1003453225631752204|resurrected, new, curious, cold|
|1003453273140633602|resurrected, new, curious, cold|
|1003453281386635271|resurrected, new, curious, cold|
|1003453356783443971|resurrected, new, curious, cold|
|1003453479206789136|resurrected, new, curious, cold|
|1003453534567407624|resurrected, new, curious, cold|
|1003453692046745605|resurrected, new, curious, cold|
|1003453723172675588|resurrected, new, curious, cold|
|1003454204951404556|resurrected, new, curious, cold|
|1003454372992000012|resurrected, new, curious, cold|
|1003454424493858820|resurrected, new, curious, cold|
+-------------------+-------------------------------+
If we focus on the first user for example, we can see the follow:
For reference check the kite records: 1003452005647130643-1564617600000
(1st of August 2019, resurrected), followed by 1003452005647130643-1564704000000
(2nd of August 2019, new)
After we do the potential user's analysis, we need to make sure that kite is working properly.
There's this bug as reported by Rami
And there are also the gaps in the data. Every user should have a entry for each day of activity until they drop off the map.
Assigning this to @ramialbatal first so he can get together any issues he was having with the current kite algorithm
So, the way we calculate new users might be the source of this false positives
A user is marked new based on the timestamp of their guid compared to the past 24 hours when the job is run
A user is marked resurrected if their bucket has activity for the first day
All the logic for these lives in UserActivityBuckets.php
@eiennohi let me brief you on this one when you're ready to work on it. Kite is best explained live.