Asked about Features for real tabular problems
How would you build features for a churn prediction model?
An answer that works
Behaviour aggregates over several windows, such as usage, logins and support tickets in the last 7, 30 and 90 days, trends between windows, tenure, plan and billing features, all computed strictly before the prediction date. Then check importance and look hard at any feature that seems too good.
Then they push further
A feature called 'account_status' is by far the most important. Concerned?
Very. It may be updated when the customer churns, which means it encodes the answer. Check when it changes relative to the churn date. If it is set at or after churn, remove it.
How do you make sure features in production match training?
Compute them with the same code in both places, ideally from a feature store with point-in-time lookups, and monitor feature distributions in production against training.
What gets marked up
Marked up for point-in-time correctness and suspicion of features that are too good. Lists of features with no mention of leakage are marked down.
Reading an answer is not being able to give one. The section that teaches this is Features for real tabular problems, 5 hours of Feature engineering.
Learn this properly