Feature Stores: When a Table and a Join Are Enough
On made-up data, a naive join scored a churn rule at 0.96; point in time, the same rule scored 0.69.
Video #159 ยท MLOps and deployment
Steps in the video
- 0:00 A naive join scored 0.96. Point in time, the same rule scored 0.69.
- 0:39 The idea
- 1:07 Setup
- 1:28 The made-up data
- 1:58 The latest table
- 2:13 The naive join
- 2:25 The point-in-time join
- 2:44 A one-line model
- 3:40 The online lookup
- 4:09 When a table and a join are enough
- 4:43 What this does not show
- 5:12 Mistakes to avoid
- 5:35 My take and next steps
- 5:58 Recap
Files
- data.py (24 lines)
- joins.py (14 lines)
- rule.py (10 lines)
- run.py (32 lines)
Tested with
Python 3.12.15
What this does not show
Made-up users with a leak I built in on purpose, one seed, a one-line rule scored on the rows it was picked on; no feature store is run, nothing is timed, and vendor claims come from their own docs.
Sources
Learning IT by Example.