- Simplifies Forecast Development
- Improves Forecast Accuracy
- Automatically Interfaces to Sales History
- Easy Management Simulation and Amendment
- Automatically Updates All Planning Modules
This module provides marketing with an automated statistical analysis and projection tool that effectively reduces the effort required to develop accurate product forecasts. Although various statistical techniques are employed, a user with little or no background in statistics can easily comprehend and utilize all capabilities of the system.
Sales History Maintenance
Up to 36 months of either demand or sales history can be maintained as a basis for forecasting. This history includes both the demand quantity and demand frequency. Demand history can be user-amended if desired. Both the actual and amended demands are maintained.
A standard interface to the Sales Analysis module is available for automatic history aggregation and update by product. If Sales Analysis is not installed, other application systems can easily update this history as well.
An item cross-reference can be defined to link the history of established products to new products if a similar demand pattern is expected. Abnormal demand filters are also available to enable the system to automatically identify and adjust abnormal occurrences of demand.
Forecast Methods
Statistical forecasting involves the use of regression analysis combined with exponential smoothing techniques. Through a series of simulations, the forecasting process examines the history for any evidence of trend and/or seasonality patterns in the data. As a result, the system selects the forecast method that best fits the time series data.
Multiple forecast methods are supported, including Average Only, Average with Trend, Average with Seasonality, Average with Trend and Seasonality, and Group Distribution to the Item Level.
Smoothing factor simulations are also performed in order to select the optimum set of smoothing factors to use when weighing new demand against established history. A separate smoothing factor is used in determining each element of the forecast which includes:
- Average Only
- Average with Trend
- Average with Seasonality
- Average with Trend and Seasonality
- Average Deviation
- Mean Absolute Deviation
- Standard Deviation
- Group Distribution to the Item Level
Override values for each element of the forecast can be specified.
Forecast Generation
A sales forecast is projected for each future period, taking the product launch and discontinuation dates into account, up to the user specified number of periods. Each element of the total forecast quantity is maintained separately by period so that the direct effects of trend, seasonality, etc. can be isolated and monitored. Each subsequent period of new history is then introduced to the system and weighed into the forecast calculations. A new, revised forecast is then generated.
Decimal accumulation routines ensure that only integer forecast values are generated. A Roll Excess option can be specified to roll any excess forecast over actual demand from the previous period into the forecast of the new first period.
Several tracking controls monitor the performance of each newly-generated forecast. If the system detects a significant change in the demand pattern for a product, it will automatically invoke a forecast method and/or smoothing factor re-simulation in order to determine whether a new method or new smoothing factors yield greater forecast accuracy. These controls can be manipulated in order to vary the sensitivity of the tests.
Parameter controls are also available to limit and diminish the effect of a linear trend over the forecast horizon.
Additional filters can be controlled to establish boundaries on the system generated forecasts.
If the computed forecast exceeds these boundary limits, then the forecast is automatically amended upward or downward. This is identified as a system amendment.
Override values for the following elements of the forecast can also be specified. Both the system values and the override values are maintained by the system or comparison purposes:
- Forecast Method
- Seasonality
- Average
- All Smoothing factors
- Trend
- Item Contribution Percentage
Manual and Firm Forecasts
Manual amendments can be applied to each period forecast if necessary. These are identified separately from system amendments. Both the system-generated forecast as well as the user-amended forecast are recorded.
Firm forecasts can also be user designated by period. This means that a forecast generation will not alter a forecast once it has been designated as firm. If desired, a system option can be set that will authorize the system to automatically firm one or more of the initial periods in the forecast horizon. This feature can be used to provide stability for near-term scheduled receipts.
Product Group Forecasts
Unique forecast groups can be assigned to items and used exclusively for forecasting purposes. Alternately, existing product group assignments (via other modules) can be utilized if desired.
Product group forecasts can be statistically-generated and then distributed to some or all of the items within the group based on the item’s group contribution percentage of demand. This can provide better forecast accuracy for those items with scanty demand history.
The contribution percentage is calculated by the system and can be amended before distributing the forecast to the items within the group. The product group forecasts can be forced to equal the sum of the forecasts for the items within the group.
On-line Forecast Simulation
After reviewing the newly-generated forecasts and forecast parameters, Sales Management can override the results and simulate the affects of their own amendments on-line, real-time without affecting the current forecasts. The system responds by showing the new forecasts compared to the existing forecasts along with an overall effectiveness measure for each. Any amendments made to the forecast and forecast parameters are stored separately from the system generated forecasts and parameters for continual comparison and review.
Multiple Forecast Iterations
The Forecast Generation program can be run as many times as management requires. Each new run takes into consideration management’s latest amendments to forecast methods, forecast parameters, etc.
In fact, the parameters used during each iteration of the Forecast Generation program can be automatically stored by the system for future consideration. This capability allows management to reload the parameters used during an earlier forecasting run prior to the next run of the Forecast Generation program.
A complete forecast history is maintained by product and period for forecast performance measurement. A forecast history extract routine is provided to periodically off load this history.
Upon management approval, the forecast projection can directly update the active forecasts that drive the various planning modules throughout the system.