Forecasting refers to the process of making statements about future occurrences where the actual outcomes have not yet been observed.
Forecasting is an inexact science that is an invaluable tool if the following is kept in mind:
- Forecasts should be tracked.
- There should be a measure of reasonableness of error.
- When actual demand exceeds the reasonableness of error, an investigation should be made to discover the cause of the error.
- If there is no apparent cause of error the method of forecasting should be reviewed to see if there is a better way to forecast.
Forecasting is done to reduce the uncertainty in the future and to help us prepare for that future in the short, medium and long term. This is done through scheduling, acquiring resources and determining resource requirements.
There are many types of forecasts, often interrelated. Broadly, forecasts are produced for situations that are quantitative, qualitative or unpredictable. For demand forecasting, we generally deal with time series (quantitative) and try find the patterns and then to predict their continuation. Such forecasts may be modified by qualitative input, such as the expected state of the economy or markets in which the business operates.
In evaluating the time series, does it exhibit a horizontal pattern, a trend (up or down), seasonal behavior, cyclic behavior, or a combination of some of these? The algorithms (logical steps and calculations used in determining the forecast) that are used are designed to address a variety of characteristics of the time series to find the best forecast.
The output from the forecast is the level of the expected demand into the future, incorporating any trends, seasonal or cyclic patterns that may be present. In addition, we need to know how reliable the forecast is. By definition, all forecasts have an error, but we need to know just how good our forecast is, because that will influence how much inventory we need to hold to provide the kind of service deemed necessary for a particular product.
The quality of the forecast can be improved by accurate record keeping both of actual sales and, if possible, unfulfilled demand. It is also important to track qualitative data that can help evaluate the forecast in the context of the markets in which the business operates.
The Inventory Forecasting system assists the enterprise in the following important areas:
- It looks at past sales and matches the algorithm which would have had the best chance of success in estimating previous sales. That algorithm is then used to predict future sales. The method described is the Competition method as it competes for the least or smallest forecast error possible. Algorithms used include:
- Moving average
- Exponential smoothing with trends
- 6 Period weighted average
- Annual seasonal profiles – unsmoothed
- Holt-Winters additive
- Holt-Winters multiplicative
- The forecast can be manually edited, where required, taking cognizance of market intelligence or correcting where the selected algorithm does not produce an adequate forecast and thereby improving accuracy and accommodating unusual demand. Thus organizations are able, through exception processes, to identify stocking items by location which require manual correction. These exceptions may include:
- A items or the company’s most important trading stock, which should be very closely monitored and managed
- Critical Security stock. Low usage items which are kept in order to fulfill customer obligations in terms of, for instance plant maintenance. Usage here may produce unusual forecasting characteristics which should be corrected
- Exceptional demand fluctuations. The cause needs to be identified and corrected.
- It records Forecast Errors to enable management to monitor and continually minimize those errors to acceptable corporate policy settings.
- It is tightly integrated into the SYSPRO transaction based system. Therefore once the forecast has been approved, it becomes a current forecast. The SYSPRO Requirements Planning module accepts the forecast which will handle the company’s replenishment calculations and processes.
- The system automatically recognizes seasonal variations and accommodates those variations.
- Large spikes in demand caused by unusual events such as export orders or stock outs can be automatically corrected by the Forecasting engine, through the use of a simple statistical outlier correction. This process is normally applied to the organization’s less important items B to D items) which are too numerous to correct manually. Unusual demand events can also be excluded manually by editing sales history.
- Selection sets can be used to segment the stock items into sensible groups for managing the forecast process. In this way different people in an organization can manage demand and supply for different sets of stock codes.
- For new, generic, superseded or substitute items, the system allows the use of a Proxy item. In such cases no (or inadequate) history is available for the forecast, so the history from the Proxy Item can be used in part or in full to predict future demand.
- Since the system is better able to predict demand it will facilitate factory scheduling. A closed loop manufacturing system is now entirely possible with continual feedback from the shop floor and the customers. Manufactures are better able to optimize their factor throughput while meeting the dynamics of the marketplace.
- Finally, the forecast engine produces a set of both tabular and graphical reports to assist in managing the forecast process. The most important of which is the ABC or Pareto analysis. This report by line item shows sales
- Cumulative Percentage of Total sales
- Percentage of Total Sales
- Sales Hits
- By Gross Profit contribution
- By Quantity
- By Cost
- By value
- On Hand value
- On hand Percentage
This report combined with the Forecast Error report focuses management attention on the most important items in the organization and assists in addressing shortfalls in the procurement process.
- In addition, Forecasting can be performed at a level above the SKU, through the use of Families and Groupings, which allows for the logical grouping of items. Items can belong either to a hierarchical family structure or to a grouping across families or both. Families and Groupings allows the user to:
- Analyze inventory movement, generate forecasts and monitor inventory parameters at levels above the item/warehouse level item/warehouse level Group items that can be dealt with in a similar manner. These groupings are particularly applicable within the Inventory Optimization module
- Aggregate the forecast into higher groups or disaggregate it back to the item/warehouse level
- Globally increase or decrease the forecast for a group of items instead of for each item/warehouse.
Forecasting is an essential part of all business planning. The Inventory Forecasting Module provides sales forecasts with simple workflow to compare and validate the forecast before approval. Forecasts are produced and reviewed at regular intervals, namely weekly, monthly, quarterly or annually for a prescribed period into the future (the forecast horizon). The optimum frequency of the forecast is dependent upon the nature of the business and the markets in which it operates, but generally more frequent forecasts allow better response to changing conditions.
Within each forecasting cycle a number of steps are followed. History must be accurately maintained and as up to date as possible. Unusual events should be excluded if they will distort the forecast. This may be done through the use of filters to eliminate outliers and by the manual exclusion of specific sales or by manual adjustments to sales or by a combination of some or all of these.
The items to be forecast should be identified and reviewed periodically. Of particular importance is to identify those items that are to be manually forecast and those that the system will forecast automatically each forecast cycle. It is also important to distinguish, and flag accordingly, those items, which will not be replenished. These could be make-to-order items or back-to-back buy-outs. It may be desirable to forecast such items but not to hold any stock of them.
The system holds two forecasts – the current forecast which is the working forecast within Requirements Planning and the draft forecast which is held in the Inventory Forecasting module until the forecast is approved and the draft then becomes the current forecast. There are two ways to produce the draft forecast, batch or manual.
The batch forecast is usually run to forecast the bulk of the items. The manual forecast process is used to deal with items that are difficult to forecast or of particular importance to the business or require the input of market intelligence to improve the quality of the forecast. The manual forecast process will produce a suggested forecast, which can then be copied to the draft for editing. The draft forecast is then compared to the current forecast to ensure that it is reasonable. Items with high forecast error values are reviewed and adjusted as necessary. The batch forecast can also be run with different parameters to compare various forecast scenarios. Once the optimum forecast is obtained it is signed off as approved. As mentioned above, the approval of a forecast makes that forecast the current forecast in Requirements Planning.
Process Flow for Inventory Forecasting
The Inventory Forecasting module integrates to the following SYSPRO modules:
The Initial History Creation program builds forecasting sales history using Inventory Movements as its source.
Once the Initial History Creation program has been run, sales forecasting data is kept current automatically by SYSPRO. As Sales Orders are written to the Inventory Movements file, the forecast history is simultaneously updated. Currently, only sales transactions are extracted from the Inventory Movement table into the Forecasting Sales History tables.
If insufficient history exists in Inventory Movements, additional history can be imported using the IOPSSL – IO forecasting sales history setup business object.
After an initial take-on of sales history movements SYSPRO will maintain the sales figures automatically.
The approved forecast from Inventory Forecasting becomes the current forecast in Requirements Planning. The draft forecast is stored in Inventory Forecasting (or Families and Groupings) while the current forecast (which can be viewed in the Manual Forecasting or Families and Groupings programs) is stored in Requirements Planning. Note that any forecast maintenance done within Requirements Planning only changes the current forecast stored there. It has no effect on the draft forecast stored in Inventory Forecasting.
Inventory Forecasting can maintain and report on both the draft and current forecasts. Changes should therefore be made in Inventory Forecasting or Families & Groupings and the changes written to Requirements Planning through the approval process.
Requirements Planning is not required for the Inventory Forecasting or Families and Groupings modules to work; however, there is no benefit to having Inventory Forecasting without Requirements Planning because nothing else in SYSPRO looks at forecasts.
Families and Groupings
This module allows forecasts to be produced at collection level. Forecast products in the Inventory Forecasting module can be used and reviewed in Families and Groupings and vice versa.
- Requirements Planning
- Sales Orders
- Families and Groupings
Inventory Forecasting Integration to other SYSPRO solutions
For more info about Syspro, you can also visit the Syspro website at Syspro