Skip to content
Predict.ai

In production

Data drift

Data drift is the inputs changing shape — new POS codes, a missing promo feed, a shifted mean. Concept drift is the relationship changing. Both make last month's champion quietly worse.

Watch the raw series, not only the error. If actuals drop to exactly zero every Sunday, you may have a file that does not land, not a demand miracle.

Concept drift is sneakier: the same promo flag no longer lifts because the market got used to it. Error goes up while the dashboard still looks 'connected.'

The fix is a cadence: refresh, re-score, and a watchdog baseline. Drift is not a once-a-year science fair.

The world moved. The model did not.

Stale forecast versus actualslast fitactuals after a break
A step-change in demand (lime) while last quarter's champion (gray) keeps drawing the old slope. That gap is staleness. Cadence and a watchdog baseline catch it before the monthly review.

Read next

Related terms

Try it on your data.

Connect an outcome and its history. Predict.ai finds the drivers, races the models, and keeps the forecast live — in the workspace, over the API, and through agents.