Skip to content
Predict.ai

Fundamentals

Granularity

Granularity is the size of each tick: hourly, daily, weekly, SKU-store versus category-region. Finer grain looks more detailed and is usually noisier. Pick the grain the planner actually uses.

You can always add daily forecasts up to a week. You cannot honestly split a weekly forecast into days without smuggling in a shape. If ops needs days, forecast days.

SKU-store daily demand is a hard problem. Category-week is easier and still useful for buying. Many teams forecast both and reconcile. That is hierarchical forecasting, not indecision.

Changing grain changes every accuracy number. A 6% WAPE on chain-week and a 40% WAPE on SKU-day can be the same system. Quote grain with the score or the score is a toy.

History · origin · horizon

Forecast origin and horizonHistory (train)Horizon (the ask)origin
The origin is 'now' for the model. Everything to the right is the horizon you asked for. Grain is how big each tick is — here, weeks.

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.