1485152 - Add `active_addons` to clients_daily to remove Longitudinal dependency for TAAR

Status: RESOLVED FIXED

(Data Platform and Tools :: General, enhancement, P2)

Description

[bugzilla]

TAAR currently has three jobs that depend on longitudinal: taar_locale and taar_similarity in python_mozetl, and addon_recommender in telemetry-batch-view. In all of these cases, the job only takes the most recent ping from a client. They all also benefit from fresher data -- that is, while longitudinal is clients from 1 sample_id over six months, if we can get an equivalent number of clients by using more sample_ids over a shorter period the recommendations would probably fare better.

Given this and the high likelihood that longitudinal will be deprecated in the near future, switching the upstream dataset seems wise. clients_daily would work pretty well for this purpose except it's missing a few fields that TAAR needs (most obviously, the full active_addons column).

I'm going to NI Victor so he can let us know what other fields the code relies on.

Comment 1

[Victor Ng [:vng] [:crankycoder]]

taar_locale could easily run off of clients_daily with the active_addons column added.

taar_similarity also needs just the active_addons column added.

AddonRecommender (for collaborative recommendations) could also be ported to use clients_daily if we had active_addons added as a column.


TADAR (formerly TAAR-lite) could also be ported if we had active_addons added.

I don't think we need any other columns added for TAAR or TAARlite other than active_addons.

Comment 3

[GitHub Bugzilla PR Linker]

Attachment: Link to GitHub pull-request: https://github.com/mozilla/telemetry-batch-view/pull/490

Comment 4

[bugzilla]

I just did a quick spot check on the last day's clients_daily run and the column is in place and the data I checked looks good