A public hospital district throughput desk receives new curated Parquet nightly and wants consistent model refresh. What automation fits?
Select an answer to reveal the explanation.
Short Explanation
New Parquet at midnight shouldn’t wait for someone to notice—schedule the nightly retrain so the hospital desk wakes up to a fresh model.
Full Explanation
A public hospital district throughput desk that receives curated Parquet nightly needs consistent refresh via scheduled or event-driven automation: nightly retrain jobs that trigger when new Parquet lands keep models aligned with arriving civic clinical-operations data without relying on human memory or email. Waiting for a manual analyst to click Run each morning introduces missed refreshes, uneven latency, and single-person failure modes on weekends and holidays. Retraining only on ad-hoc email requests has no reliable cadence and fails when the requester is out or the inbox is noisy. Disabling jobs when new data arrives intentionally freezes the model as the hospital’s operational distribution drifts, which is the opposite of a consistent refresh strategy. Exam caveat: municipal nightly landings map to scheduled or event-driven retrain automation, not manual clicks or email-gated runs. Operational check: wire the Parquet landing event or cron to the retrain workflow with retries, success metrics, and on-call alerts so the throughput desk can trust that morning scores reflect last night’s curated extract. In municipal practice for this NCA-ADS scenario (item q095), document the decision, retain evidence for auditors, and re-validate holdout civic metrics after any change so operational owners can defend the choice under scrutiny.