This main program generates a forecast, order quantity, safety stock and related replenish
measures for each part and location. It identifies the parts with low (on-hand plus on-order)
inventory; with high on-order inventory; and with surplus inventory. Summary and detail
reports are printed allowing management to monitor the inventory on all parts. The key results
are stored in the database for use by each of the remaining ten routines.
Forecast-Replenish conducts a statistical analysis of up to 24 months of demand history for
each part to determine the best forecast model from either a horizontal, trend, seasonal or
promotion model. Outliers are detected and processed in accordance with a user- selected
parameter. If fiscal months in a quarter are of the 4, 4, 5 week type, then the forecasts are
The selected model is used to generate 12 future monthly forecasts. Forecast-Replenish's
flexibility is evident in the three ways it allows the forecasts to be altered by a user
- Directly adjust each of the 12 months forecasts through 12 monthly percentages.
- Manual forecasts may be entered.
- The statistically generated forecasts may be altered if there are orders for future months
(advance demand) that are known in the present month.
Forecast-Replenish does not use the "trial-and-error" methods that are in common use
today. These trial and error methods merely seek the forecast model or parameters from a
moderate set of models and parameter settings, that yields the best using only a portion of the
history demands. These trial-and-error approaches are very sensitive to lumpy swings in
monthly demands and the forecast models often alternate from one month's run to the next; and
thereby the forecasts are not statistically consistent.
Additionally, the Weekly routine converts the 12 monthly forecasts generated from Forecast-
Replenish to 52 weekly forecasts. The routine applies for calendar months and for fiscal (445)