Data processing: generic control systems or specific application – Specific application – apparatus or process – Product assembly or manufacturing
Reexamination Certificate
2000-09-15
2003-08-26
Gordon, Paul P. (Department: 2857)
Data processing: generic control systems or specific application
Specific application, apparatus or process
Product assembly or manufacturing
C700S052000, C700S100000, C700S106000, C705S002000, C705S028000
Reexamination Certificate
active
06611726
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a new method for calculating time series forecasting parameters that includes looking forward to improve forecasting accuracy for supply chain needs beyond the next time period. In particular, the invention is directed to a method for estimating the parameters of a selected forecasting method such that the forecast is not necessarily most accurate in the next, upcoming time period but is instead optimized to give a more accurate forecast for some specific, user-selected future time period. The method is also applicable to estimate time series forecasting parameters when it is necessary to optimize a time series forecast for two or more future time intervals combined.
BACKGROUND OF THE INVENTION
Organizations need to carefully plan and be prepared for their future business in order to be successful, and the required planning includes understanding supply chain complexities and predicting future behaviors and values of planning variables, such as product demand. For example, organizations must plan to have sufficient product available to meet demand while not having excess product that may quickly become “stale” or out dated. Outdated inventory is a serious and potentially costly problem for companies that manufacture and sell high technology and other products with shorter life spans.
One measure of organizational success used by organizations for planning purposes is revenue. Consequently, as a part of planning, organizations need to accurately estimate future income as well as carefully plan future expenses in order to calculate expected revenues. One method used to calculate expected revenues is to estimate future income from historical demand or sales information and then budget expenses from the estimate of future income. However, this method does not address the complexities of supply chains and the impact supplier characteristics can have on providing components that are later utilized to manufacture products to meet forecasted demand.
Another method to calculate expected revenues is to estimate both future income and future expenses from historical information and then evaluate the reasonableness of the estimated expenses to determine if the forecasted expenses need adjusting. Both of these approaches allow an organization to determine a relationship between income, expenses, and revenues, and of course, there are numerous other methods that are utilized by organizations for strategic planning.
Estimating income and other planning variables from past or historical information is always necessary regardless of which of the planning approaches is selected. Typically, some form of forecasting process uses the historical information to provide these estimations. As a specific example, organizations estimate future income by forecasting the future demand for each individual product. This approach is used since each product is likely to produce a different income per unit.
In addition to providing information about estimated income for the planning group of the organization, the forecasted demand provides information that is relatively consistent across the organization for other groups within the organization to use. For example, if the organization manufactures a product, the forecasted demand for each product is often used to determine the effect on the organization's ability to manufacture sufficient products. The manufacturing group needs to determine if their manufacturing capabilities are adequate and if they have sufficient inventories of component parts to produce the potential product demand. Without adequate manufacturing capability or sufficient inventory, the organization may not be able to manufacture enough products to produce planned incomes. This would reduce income and adversely affect expected revenues.
The manufacturing group also relies on forecasted demand to allocate its resources efficiently, balance workload against the forecasted demand, and plan their operations to meet the needs the product demand places on them. Without this manufacturing planning, manufacturing may not be able to produce products in a timely manner, which would limit the availability of the product for the customer when the customer wants to purchase the product. This would also decrease sales and reduce income and adversely affect expected revenues.
Manufacturing organizations occasionally use just-in-time manufacturing. This manufacturing approach minimizes excess inventory by providing the component parts to assemble a product just in time for the assembly process. This reduces or even eliminates any component part inventory, which minimizes the risk of manufacturing components becoming outdated. However, even organizations that use the just-in-time approach must accurately determine how many and when each component will be needed. Such a determination again involves forecasting or predicting future demand and/or need that will exist in the next time period and, more likely, in the next several time periods to allow components to be ordered and delivered by a supplier.
A lead time problem often occurs when ordering component parts. Some suppliers can provide component parts with little or no notice, i.e., no or little lead time. Other component part suppliers may require the manufacturer to place orders for components several months in advance, even as much as six months in advance. The reasons behind this requirement may range from the length of time needed to make the component part to transportation delays for shipping the parts from manufacturing sites abroad and to a simple need to be able to plan their own future manufacturing effort. The varying lead-time problem is made even more significant if these long lead-time components are also very expensive.
For example, the manufacturer of some expensive components, such as computer components, may require orders be placed three months in advance and these orders may also be non-cancelable. Consequently, accurately forecasting how many of these components a company will need in three months or a time period coinciding with a supplier's required lead time, is a very important part of effective organization planning, such as inventory management. Further in this regard, manufacturing organizations do not want to inaccurately order expensive components because having too few components will result in lost income from the product but having too many components will result in unnecessary expenses. Also, sometimes these expensive components have a short lifetime of usefulness and having excess inventory of outdated components may lead to wasted expense if these components need to be discarded.
Clearly, accurate forecasting is important to the success of a business or other organization, but a perfectly accurate forecast is not possible, even with existing forecasting software products and computer-implemented computation methods. Generally, the process of forecasting product demand includes accessing historical product demand data, analyzing the data to determine if they are suitable to use for forecasting, and then processing the historical demand data to produce demand forecasts for the future.
Existing forecasting typically begins with accessing historical product demand data. Organizations store their historical data in a variety of data sources. The data necessary for forecasting may be stored in spreadsheets, databases, data marts, or data warehouses. This data may be retrieved using OLAP (on-line analytical processing) engines, MOLAP (multiple OLAP) engines, ROLAP (relational OLAP) engines, Open Data Base Connectivity (ODBC), Java's JDBC data connectivity and other methods that allow for data mining and analysis. Once historical demand data has been retrieved, data analysis determines if there exists data anomalies such as outliers, missing data, and the like that require adjustments to the historical data to smooth or remove these anomalies. Various smoothing methods in use today include setting missing data to zero, averaging neighboring data to correc
Gordon Paul P.
Hogan & Hartson LLP
Lembke Kent A.
LandOfFree
Method for determining optimal time series forecasting... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for determining optimal time series forecasting..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for determining optimal time series forecasting... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3114850