Modified, redesigned in version 7.5.2
The large sales forecast filter is a statistical process that compares sales orders being entered as current demand against a historical sample. Sales orders with a quantity beyond a specified upper statistical limit are subject to filtering.
Forecast filtering is applied to items by forecast group. Forecast group settings are defined in program FCGM, Forecast Group Maintenance.
If filtering is active for a forecast group, demand is loaded from sales orders or sales order shipments for the forecast item, depending on the forecast settings for the item on the item planning record in program ICPM, Inventory Planning Data Maintenance.
* The settings for a specific plant on the Plant Tab in program ICPM, Inventory Planning Data Maintenance, take precedent over the settings for the enterprise on the Default ME Tab.
The demand quantity and the values applied by the filtering process are recorded for each demand type. These values are used at a future time in order to calculate average order quantities.
The average demand quantity and standard deviation of the demand quantity are calculated for all demand records which fall within the sample interval between the forecast calculation date and the previous value for the large sales average days.
* The average demand per forecast is calculated as the total sample demand * forecast days / average days.
For each demand record in the current forecast period, the Z factor is calculated for the demand quantity. The Z factor calculation is:
Z= (order quantity - average order quantity)/standard deviation.
Z factors that are above the Z limit for the forecast group are subject to filtering unless the demand quantity is below the minimum quantity for the forecast group.
If the demand record is subject to filtering, the quantity will be limited to the limit percent * average demand per forecast as described above.
If the number of demand records supplied to the filtering procedure is greater than the minimum population defined for the forecast group, then the filtered quantity is used as the current demand. Otherwise, the raw demand is used.
To overcome problems that result from infrequent demand, a proxy demand record for a quantity of 0 is created for forecast items with no current demand. These records are distinguished by having the order number correspond to the calculation date and a line number of 0. These records participate in future calculations to give presence in the population for periods that have no demand.
For forecast transfers with program FCCX, Write Statistical Forecast Data to MPS, if the FCCX_AUTO_FC_ACTIVATE parameter in program MURM, Initialization Parameter Maintenance, is set to Y, the items that have the Statistical Forecast flag set to Track forecasts in the item planning programs will have the setting changed to Yes if the Filter Enabled flag is set to on for the item's forecast group in program FCGM, Forecast Group Maintenance and the sample population exceeds the minimum population for the forecast group. This allows the forecasting for new items to be set to Track initially, and have the forecasting be automatically activated when enough data exists.