How do you safely backfill features / labels historically?

hard

Answer

$(\mathrm{backfill}_{\mathrm{YYYYMMDD}})$
$(2) Validate output matches sample of production for overlap dates.$
$(3) Batch by date partition to bound resource use.$
$(4) Coordinate with online store rebuild (avoid inconsistency).$
$(5) Communicate to downstream consumers.$
$(6) Test model retrained on backfilled data on golden set before switching.$
$Anti-pattern: overwrite production data before validation — irreversible if wrong.$

Check yourself — multiple choice

Just overwrite
Isolate to non-prod table + validate overlap dates + batch by partition + coordinate online store + comm to consumers + test retrained model; never overwrite prod first
Random
Not real

Backfill: isolate + validate + batch + coordinate + test before swap.

#pipeline

Practise MLOps & Data Quality

215 interview questions in this topic.

Read the lesson Flashcards Take the quiz

Related questions