Background: Forecast Calculation

Use the Forecast Calculation (MFC) program to calculate forecast demand for all items. This program should be run at regular intervals for reliable results. See Road Map for work flow.

Forecast Calculation bases the calculation on the Accum demand and New forecast fields in the Branch Item Maintenance (IMB) program. The program then updates the New forecast, MAD, and Trip signal fields and sets the Accum demand field to zero.

This topic has these subtopics:

New Forecast
MAD, Trip Signal, and Safety Stock

Forecast Calculation Report

New Forecast

The program calculates the New forecast field as:

Previous new forecast + Smoothing factor x (Actual demand - Previous new forecast)

  • The previous forecast is entry in the New forecast field for the preceding period.

  • The smoothing factor is your entry for the Enter new factor Program option.

  • Actual demand includes all orders entered through Sales Order Maintenance (OE) and One-Step Invoice Entry (RE) during the preceding period.

The forecast error is Actual demand - Previous forecast. For example, if the actual demand is 100 and the last new forecast was 80, the forecast error is 20.

This method of calculating forecasts is called exponential smoothing.

If an item has no master schedule, the Material Requirements Planning (MRP) program uses the New forecast value for the planning of material requirements. The Forecast Update (MFU) program also uses it to update an item's forecast, which the Forecast Maintenance (MSF) program displays.

You should run the Forecast Calculation program at consistent intervals. Actual demand is the demand registered since the last time you ran the program. This period should equal the period covered by the Last new forecast. If it does not, the forecast error will be inaccurate and the total calculation invalid.

The result of running the Forecast Calculation program appears in the New forecast field in the Branch Item Maintenance (IMB) program. Old forecast displays the previous value from the New forecast field.

See Overview: Forecasting for other forecasting methods.

MAD, Trip Signal, and Safety Stock

Forecast Update also calculates and updates these fields in the Branch Item Maintenance (IMB) program.

Field Name

Description

MAD

Mean Absolute Deviation is the absolute average of the forecast error over a number of planning periods. MAD indicates the accuracy of  previous forecasting.

Forecast error = ABS (Actual demand - Previous new forecast)

New MAD = Last MAD + (New factor x (Forecast error - Last MAD))

Example

Last MAD was 63
Last new forecast was 2063
Demand was 2500

2500-2063 is 437 or the forecast error

New MAD is 137 based on 63 + (.2 * (437-63))

        63 + (.2 * 374)
        63 +74
        137

Trip signal

The program registers a value of 1 every time the forecast error for an item exceeds three times the MAD (Mean Absolute Deviation) value. When the Trip signal reaches 3, the item appears on a trip list that this program prints. This list warns the planner of excessive unplanned demand or forecast variance for the item.

Safety stock

If you selected the Safety stk upd? field in the Branch Item Maintenance (IMB) program, the program also updates the Safety stock field in that program. The Mean Absolute Deviation (MAD) represents the uncertainty of the forecast, and the program uses its value to compute the amount of safety stock that is appropriate for each item. Forecast Calculation computes safety stock as 1.6 x MAD, which yields a 90 percent service level.

Forecast Calculation Report

The report includes this information:

  • Branch

  • Item number

  • Item description

  • Trip signal

  • Old forecast: Quantity before recalculation

  • New forecast

  • Accumulated demand

  • Mean absolute deviation or average forecast error

If you selected the Print only totals? Program option, these totals that print at the end of the report are all that prints:

  • Items forecasted up

  • Items forecasted down

  • Items with no change

  • Percent of total: Items forecasted up, down or with no change / Total number of items

  • Total number of items: Items forecasted up + Items forecasted down + Items with no change