Appearance
Demand forecasts
Also known as
- demand forecast
- statistical forecast
- demand planning
- exponential smoothing
- Holt-Winters
- moving average
- MAPE
- backtest
- forecast override
- MRP
The forecast list shows one row per item × warehouse (a "series") with the method that forecasts it, how accurate that method was in the backtest and the quantities it expects. Planners use it to find series that need a closer look or an override. The final forecast (the override when there is one, else the selected method) feeds MRP in Manufacturing, the reorder run, DRP, the forecast shared with suppliers and the S&OP plan.
Find a series that needs attention
- Open Supply chainDemand forecasts.
- Use the filters: Warehouse, Method, Overridden, Automatic method and Backtest MAPE (for example, MAPE above 30 to find the poorly forecast series).
- Sort by Backtest MAPE or Next 3 months.
- Select a row to open the forecast of that item and warehouse, where you can override months or fix the method.
The links under the title open Forecast accuracy, Forecast runs and Demand history.
Columns
| Column | Meaning |
|---|---|
| Item, Warehouse | The series. |
| Method | The method whose numbers are used. fixed means a planner chose it; it survives later runs. |
| Backtest MAPE | Mean absolute percentage error of the selected method over the backtest months. |
| MAPE · Moving average / Exponential smoothing / Holt-Winters (seasonal) | The backtest MAPE of each method, to compare. |
| Average per month | Average monthly demand in the history window. |
| This month, Next 3 months | Final forecast of the current month and of the next three. |
| Overrides | Months a planner overrode. |
| Last run | When a forecast run last computed the series. |
Recompute the forecast
Select Run forecast (needs Permission: supply_chain.forecast.run). It opens the New forecast run dialog. The monthly job also recomputes every series on the 1st of the month when Recompute the forecast on the first of every month is on in Forecasting settings.
Where the demand comes from
Demand history is the quantity shipped to customers from Sales, per month, plus the demand history you import from earlier systems. Each run fits every enabled method, backtests it month by month and keeps the method with the lowest MAPE (unless a planner fixed one).
