WO2015065343A1

WO2015065343A1 - Optimizing a forecast with a disaggregation ratio at a forecast level

Info

Publication number: WO2015065343A1
Application number: PCT/US2013/067339
Authority: WO
Inventors: Pitchu Kumar ESWARAMURTHY; Jaishankar BHASKARAN NAIR; Abinesh BALASUBRAMANIAN; Arun Kumar VADHYAR AMBIKAPATHI
Original assignee: Hewlett-Packard Development Company, L.P.
Priority date: 2013-10-29
Filing date: 2013-10-29
Publication date: 2015-05-07
Also published as: EP3063681A4; EP3063681A1; CN105874459A

Abstract

Optimizing a forecast with a disaggregation ratio at a forecast level includes determining product dimensions and geography dimensions to be evaluated in a forecast, evaluating combinations of the product dimensions and the geography dimensions, and optimizing a disaggregation ratio to prepare the forecast at a forecast level.

Description

OPTIMIZING A FORECAST WITH A DISAGGREGATION RATIO

AT A FORECAST LEVEL

BACKGROUND

[0001] Forecasting is the process of estimation a future event. In business, forecasting is used to prepare a forecast for business plans, investment decisions, operational plans, supply chain planning, and other business decisions. Further, the accuracy of a forecast is dependent upon how information is used in preparing the forecast and a forecast level chosen for the forecast.

BRIEF DESCRIPTION OF THE DRAWINGS

[0002] The accompanying drawings illustrate various examples of the principles described herein and are a part of the specification. The examples do not limit the scope of the claims.

[0003] Fig. 1 is a diagram of an example of an optimizing system, according to one example of principles described herein.

[0004] Fig. 2 is a diagram of an example of an optimizing

environment, according to one example of principles described herein.

[0005] Fig. 3 is a diagram of an example of a forecast, according to one example of principles described herein.

[0006] Fig. 4 is a diagram of an example of a forecast, according to one example of principles described herein.

[0007] Fig. 5 is a flowchart of an example of a method for optimizing a forecast with a disaggregation ratio, according to one example of principles described herein. [0008] Fig. 6 is a flowchart of an example of a method for optimizing a forecast with a disaggregation ratio, according to one example of principles described herein.

[0009] Fig. 7 is a diagram of an example of an optimizing system, according to one example of principles described herein.

[0010] Fig. 8 is a diagram of an example of an optimizing system, according to one example of principles described herein.

[0011] Throughout the drawings, identical reference numbers designate similar, but not necessarily identical, elements.

DETAILED DESCRIPTION

[0012] To determine the accuracy of a forecast, a forecaster analyzes the history of a business's products. The history of a business's products may include how many of the business's products were sold, where the business's products were sold. The more information that is available to the forecaster, the more accurate the forecast may be.

[0013] However, with time, changing market conditions and

proliferations of new product lines, a forecast level may not be valid after a period of time has passed. Further, due to the intense manual efforts used to analyze a business's history, a forecast level is rarely re-evaluated and remains static throughout a period of time. As a result, this can significantly affect the accuracy of the forecast with downstream implications in supply chain and inventory management.

[0014] The principles described herein include a method for optimizing a forecast with a disaggregation ratio at a forecast level. Such a method includes determining product dimensions and geography dimensions to be evaluated in a forecast, evaluating combinations of the product dimensions and the geography dimensions, and optimizing a disaggregation ratio to prepare the forecast at a forecast level. Such a method allows all possible combinations of product dimensions and geography dimensions to be evaluated. As a result, an optimized disaggregation ratio is produced to ensure the forecast is accurate at the forecast level.

[0015] Further, the method can include calculating an optimal disaggregation path for the product dimensions and the geography dimensions based on a minimum error forecast. More detail about his method will be described in more detail below.

[0016] A forecast may be statements about events whose actual outcomes have not yet been observed. In one example, a forecast may be a statement about estimations of an event such as investment decisions, operational plans, supply chain, other events, or combinations thereof.

[0017] A product dimension may be a level at which a product may be evaluated. For example, a product dimension may be evaluated as a whole business unit, including all the products a business manufactures. Further, the product dimension may be evaluated as product categories. Still further, the product dimension may be evaluated as individual products.

[0018] A geography dimension may be a level at which a geography may be evaluated. For example, a geography dimension may be evaluated as a region, a market segment, a sub region, a country, or combinations thereof. Further, a geography dimension may include a historical value. In this example, a historical value includes how many products were sold, where the products were sold.

[0019] A time dimension may be a level of time at which a forecast is evaluated. In one example, a time dimension may be days, weeks, months, years, other time dimensions, or combinations thereof.

[0020] A disaggregation ratio may include a relationship between two or more product dimensions, geography dimension, time dimensions, or combinations thereof. Further a disaggregation ratio may be optimized to ensure an accurate forecast is produced.

[0021] A forecast level may be a level at which a forecast is prepared. In one example, a forecast level may a stock keeping unit (SKU) level. As a result, a forecast is prepared at a SKU level. In another example, a forecast level may a category level. As a result, a forecast is prepared at a category level. In yet another example, a forecast level may an aggregate level. As a result, a forecast is prepared at an aggregate level. Further, in one example, a user may select the forecast level for preparing the forecast. As a result, a forecast may be prepared at a desired forecast level.

[0022] Further, as used in the present specification and in the appended claims, the term "a number of or similar language is meant to be understood broadly as any positive number comprising 1 to infinity; zero not being a number, but the absence of a number.

[0023] In the following description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough

understanding of the present systems and methods. It will be apparent, however, to one skilled in the art that the present apparatus, systems, and methods may be practiced without these specific details. Reference in the specification to "an example" or similar language means that a particular feature, structure, or characteristic described in connection with that example is included as described, but may not be included in other examples.

[0024] Fig. 1 is a diagram of an example of an optimizing system, according to one example of principles described herein. As will be described below, an optimizing system is in communication with a network to determine product dimensions and geography dimensions to be evaluated in a forecast. The optimizing system evaluates combinations of the product dimensions and the geography dimensions. The optimizing system further optimizes a disaggregation ratio to prepare the forecast at a forecast level. Such a method allows all possible combinations of product dimensions and geography dimensions to be evaluated. As a result, an optimized disaggregation ratio is produced to ensure an optimal forecast at a forecast level.

[0025] In one example, the system (100) includes a database (1 12). As will be described in other parts of this specification, the database (102) includes a business warehouse database, an advanced planner and optimizer database, a product database, or combinations thereof. In this example, the database (1 12) includes information about product dimensions, geography dimensions, time dimensions, historical values, or combinations thereof. In keeping with the given example, the database (1 12) is in communication with an optimizing system (108) over a network (106) to use the information from the database (1 12) to prepare a forecast at a forecast level.

[0026] The system (100) further includes an optimizing system (108). In keeping with the given example, the optimizing system (108) determines product dimensions and geography dimensions to be evaluated in a forecast. As mentioned above, a forecast may be statements about events whose actual outcomes have not yet been observed. In one example, a forecast may be a statement about estimations of an event such as investment decisions, operational plans, supply chain, other events, or combinations thereof.

[0027] Further, as mentioned above, a product dimension may be a level at which a product may be evaluated. For example, a product dimension may be evaluated as a whole business unit, including all the products a business manufactures. Further, the product dimension may be evaluated as product categories. Still further, the product dimension may be evaluated as individual products. In keeping with the given example, a geography dimension may be a level at which a geography may be evaluated. For example, a geography dimension may be evaluated as a region, a market segment, a sub region, a country, or combinations thereof. Further, a geography dimension may include a historical value. In this example, a historical value includes how many products were sold, where the products were sold, and the time it took to sell the products based on a geography dimension. Further, a historical value may include historical values for a region, historical values for a market segment, historical values for a sub region, historical values for a country, or combinations thereof. More information about determining product dimensions and geography dimensions to be evaluated in a forecast will be described in other parts of this specification.

[0028] The optimizing system (108) further evaluates combinations of the product dimensions and the geography dimensions. For example, a product dimension may include three individual products, P1 , P2, and P3. Further, a geography dimension may include three geography dimensions, G1 , G2, and G3. In this example, the combinations of the product dimensions and the geography dimensions may include nine possible combinations. For example, P1 and G1 , P1 and G2, P1 and G3, P2 and G1 , P2 and G2, P3 and G3, P3 and G1 , P3 and G2, P3 and G3. As will be described in other parts of this

specification, each combination is evaluated for preparing a forecast. More information about evaluating combinations of the product dimensions and the geography dimensions will be described in other parts of this specification.

[0029] The optimizing system (108) further optimizes a disaggregation ratio to prepare the forecast at a forecast level. In one example, the

disaggregation ratio is optimized to allow a forecaster to determine which product dimension and geography dimension yield the most favorable results for a forecast. Further, a user may select the forecast level for preparing the forecast. As a result, a forecast may be prepared at a desired forecast level. As a result, an optimized disaggregation ratio is produced to ensure the forecast is accurate at a forecast level. More information about optimizing a

disaggregation ratio to prepare the forecast will be described in other parts of this specification.

[0030] The system (100) further includes a user device (102) with a display (104). In this example, the optimizing system (108) presents to a user, such as a forecaster, a forecast via the user device (1 12). As will be described in other parts of this specification, the forecast may be in the form of a

spreadsheet to allow the forecaster to easily interpret the forecast.

[0031] While this example has been described with reference to the optimizing system being located over the network, the optimizing system may be located in any appropriate location according to the principles described herein. For example, the optimizing system may be located in a user device.

[0032] Fig. 2 is a diagram of an example of an optimizing

environment, according to one example of principles described herein. As mentioned above, an optimizing system is used to determine product

dimensions and geography dimensions to be evaluated in a forecast. The optimizing system evaluates combinations of the product dimensions and the geography dimensions. The optimizing system further optimizes a

disaggregation ratio to prepare the forecast at a forecast level. Such a method allows all possible combinations of product dimensions and geography dimensions to be evaluated. As a result, an optimized disaggregation ratio is produced to ensure an optimal forecast.

[0033] In the example of Fig. 2, the environment (200) includes a number of databases (232). In this example, the databases (232) include a business warehouse (BW) database (202), an advanced planner and optimizer (APO) database (204), and a product database (206).

[0034] In one example, the BW database (202) contains information that includes information about an underlying data warehouse area. In one example, the data warehouse area is responsible for storing information in various types of structures such as data store objects, Info object, other structures, or combinations thereof. In keeping with the given example, the BW database (202) is in communication with a product database (206). As a result, the information stored in the BW database (202) may be made accessible to the product database (206). More information about the product database (206) will be described below.

[0035] As mentioned above, the environment (200) includes an APO database (204). The APO database (204) may include a planning tool which is used to plan and optimize supply chain processes by making use of various modules. In one example, the modules may include demand planning, supply network planning (SNP), other modules, or combinations thereof.

[0036] In this example, the demand planning is a set of functionalities around demand management, statistical forecasting, promotion and life-cycle planning processes. Further, the demand planning is an integral part of a business's sales and operations planning process.

[0037] In keeping with the given example, the SNP is a module in APO that orchestrates aggregated production and distribution planning across locations in a supply chain. In one example, the SNP uses a number of tools for planning the production & distribution across the various locations in the supply network. In keeping with the given example, the APO database (204) is in communication with a product database (206). As a result, the information stored in the APO database (202) may be made accessible to the product database (206).

[0038] As mentioned above, the environment (200) includes a product database (206). In one example, the product database (206) includes information about actual shipments of products. For example, what the product is, how many products were shipped, where the products are shipped, and who bought the products. In this example, the information stored in the product database (206) may be made available to an optimizing system (218).

[0039] In this example, the optimizing system (218) includes a data builder engine (208), an identifying engine (214), and an optimizing engine (216). In one example, the data builder engine (208) receives information from the product database (208). In keeping with the given example, the data builder engine (208) includes an outlier correction engine (210) and a like product mapping engine (212). In this example, the outlier correction engine (210) to correct outlier's. In one example, outliers can occur by chance in a distribution or that the population has a heavy-tailed distribution. In one example, data points will be further away from the sample mean than is deemed reasonable. Outlier points can therefore indicate faulty data, erroneous procedures, or areas where a certain theory might not be valid. As a result, the outlier correction engine (210) corrects such error in the data.

[0040] As mentioned above, the data builder engine (208) includes a like product mapping engine (212). In this example, like product mapping engine (212) maps products that are similar. For example, product X may be similar to product Y in the fact that both product X and product Y are the same product but differ in color. As a result, the like product mapping engine (212) maps product X and product Y as being similar.

[0041] Further, the data builder engine (208) may determine product dimensions and geography dimensions. Further, the data builder engine (208) may further evaluate combinations of product dimensions and geography dimensions. More information about evaluating combinations of product dimensions and geography dimensions will be described in later parts of this specification [0042] As mentioned above, the optimizing system (218) further includes an identification engine (214). The identification engine (214) identifies the best possible path for every product dimension and geography dimension combination. Further, the identification engine (214) may compare results for every product dimension and geography dimension combination to populate an error basis for determining the best possible path for every product dimension and geography dimension combination. More information about identifying the best possible path for every product dimension and geography dimension combination will be described in later parts of this specification.

[0043] As mentioned above, the optimizing system (218) includes an optimizing engine (216). In one example, the optimizing engine (216) provides optimized disaggregated ratios at a forecast level. Further, the optimized disaggregated ratios can be used as an input for the APO database (204).

More information about the optimized disaggregated ratios at a forecast level will be described in later parts of this specification.

[0044] The environment (200) can further include a results engine (224). In this example the results engine (224) includes a disaggregation error summary (220), a forecast level estimation (222) and an optimized

disaggregation ratio (226). In one example, the disaggregation error summary (220) includes an identified best possible disaggregation methodology. Further, the forecast level estimation (222) may include results for a minimum error forecast. Still further, the optimized disaggregation ratio (226) may include result for and optimized disaggregation ratio at a forecast level. More information about the disaggregation methodology, minimum error forecast, and optimized disaggregation ratio will be described in later parts of this

specification.

[0045] Additionally, the results for the disaggregation methodology, minimum error forecast, and optimized disaggregation ratio at a forecast level are sent to a user device (228). In this example, the user device (228) includes a display (230) to display the results for the disaggregation methodology, minimum error forecast, and optimized disaggregation ratio at a forecast level to a forecaster. [0046] Fig. 3 is a diagram of an example of a forecast, according to one example of principles described herein. As mentioned above, the optimizing system includes an identification engine. The identification engine identifies the best possible path for every product dimension and geography dimension combination. Further, the optimizing engine optimizes a

disaggregation ratio at a forecast level. As will be described below, the results of an optimized disaggregation ratio at the forecast level are presented to a forecaster as a forecast.

[0047] In the example, of Fig. 3, a display (302) may display the results of the optimizing system. In this example, the results of the optimizing engine may include displaying SKU (304), countries (306), optimized profiles (308), and optimized disaggregation ratios (310).

[0048] In this example, the SKUs (304) may include three SKUs. For example, stock keeping unit one (304-1 ), stock keeping unit two (304-2), and stock keeping unit three (304-3). In this example, the stock keeping units (304) may be a distinct item, such as a product or service, as it is offered for sale that embodies all attributes associated with the item and that distinguish it from all other items. In one example, for a product, these attributes include, but are not limited to, manufacturer, product description, material, size, color, packaging, and warranty terms. In another example, the SKUs (304) may include a specific identification number for a product that is distinct from all other products using SKUs.

[0049] In this example, the SKUs (304) may be identified with a geography dimension (306). For example, stock keeping unit one (304-1 ) is identified with geography dimension one (306-1 ), stock keeping unit two (304-2) is identified with geography dimension two (306-2), and stock keeping unit three (304-3) is identified with geography dimension three (306-3). In this example the geography dimension (306) may be a region, a market segment, a sub region, a country, or combinations thereof.

[0050] Further, the SKUs (304) may be identified with optimized profiles (308). In this example, the optimized profile (308) may include the combinations of the product dimensions and the geography dimensions. In keeping with the given example, stock keeping unit one (304-1 ) is identified with optimized profile one (308-1 ), stock keeping unit two (304-2) is identified with optimized profile two (308-2), and stock keeping unit three (304-3) is identified with optimized profile three (308-3).

[0051] Further, the SKUs (304) may be identified with an optimized disaggregation ratio (310). For example, stock keeping unit one (304-1 ) is identified with optimized disaggregation ratio one (310-1 ), stock keeping unit two (304-2) is identified with optimized disaggregation ratio two (310-2), and stock keeping unit three (304-3) is identified with optimized disaggregation ratio three (310-3). In this example, the optimized disaggregation ratio (310) may be displayed as percentages. For example, optimized disaggregation ratio one (310-1 ) may be thirty percent, optimized disaggregation ratio two (310-2) may be twenty percent, and optimized disaggregation ratio three (310-3) may be fifty percent. As a result, the forecast (300) allows a forecaster to determine specific product dimensions to be sent to specific geography dimensions according to an optimized disaggregation ratio.

[0052] Fig. 4 is a diagram of an example of a forecast, according to one example of principles described herein. As mentioned above, an

identification engine identifies the best possible path for every product dimension and geography dimension combination. Further, an optimizing engine optimizes a disaggregation ratio according to a minimum error forecast. As will be described below, the results of the minimum error forecast are presented to a forecaster as a forecast.

[0053] In the example, of Fig. 4, a display (302) may display the results of the optimizing system. In this example, the results of the optimizing system may include time dimensions (404), optimized profiles (406), data types (408), and minimum error forecasts (410). In one example, a time dimension may be a moving average. In one example, a moving average is a filter used to analyze a set of data points by creating a series of averages of different subsets of the full data set. Further, moving averages are a set of numbers, each of which is the average of the corresponding subset of a larger set of datum points. A moving average may further use unequal weights for each datum value in the subset to emphasize particular values in the subset. In one example, a moving average may be a three month moving average, a six month moving average, a nine month moving average, a twelve month moving average, or other moving averages.

[0054] In this example, the time dimensions (404) may include three moving averages. Time dimension one (404-1 ), time dimension two (404-2), and time dimension three (404-3). In this example, the moving averages (404) may be a three month moving average, a six month moving average, and a nine month moving average respectively. In one example, the time dimensions (404) may be identified with an optimized profile (406). For example, time dimension one (404-1 ) is identified with optimized profile one (406-1 ), time dimension two (404-2) is identified with optimized profile two (406-2), time dimension three (404-3) is identified with optimized profile three (406-3). In this example, the optimized profiles (406) include all the possible combinations of the product dimensions and the geography dimensions.

[0055] In one example, the time dimensions (404) may be identified with a data type (408). For example, time dimension one (404-1 ) is identified with data type one (408-1 ), time dimension two (404-2) is identified with data type two (408-2), time dimension three (404-3) is identified with data type three (408-3). In this example, the data types (408) may indicate that the forecast (400) includes an error comparison.

[0056] In keeping with the given example, the time dimensions (404) may be identified with a minimum forecast error (410). For example, time dimension one (404-1 ) is identified with minimum forecast error one (410-1 ), time dimension two (404-2) is identified with minimum forecast error two (410- 2), time dimension three (404-3) is identified with minimum forecast error three (410-3). In this example, minimum forecast error one (410-1 ) may be ninety percent, minimum forecast error two (410-2) may be three-hundred and twenty percent, and minimum forecast error three (410-3) may be one-hundred and fifty percent. As a result, the forecast (300) allows a forecaster to determine specific product dimensions to be sent to specific geography dimensions based on a minimum forecast error. More information about the minimum forecast error will be described in detail in later parts of this specification.

[0057] Fig. 5 is a flowchart of an example of a method for optimizing a forecast with a disaggregation ratio at a forecast level, according to one example of principles described herein. In this example, the method (500) includes determining (501 ) product dimensions and geography dimensions to be evaluated in a forecast, evaluating (502) combinations of the product dimensions and the geography dimensions, and optimizing (503) a

disaggregation ratio to prepare the forecast at a forecast level.

[0058] As mentioned about, the method (500) includes determining (501 ) product dimensions and geography dimensions to be evaluated in a forecast. In one example, a product dimension may include three dimensions, P1 , P2, and P3. Further, a geography dimension may include three dimensions, G1 , G2, and G3.

[0059] Further the method (500) includes evaluating (502)

combinations of the product dimensions and the geography dimensions. In this example, the combinations of the product dimensions and the geography dimensions may include nine possible combinations. For example, P1 and G1 , P1 and G2, P1 and G3, P2 and G1 , P2 and G2, P3 and G3, P3 and G1 , P3 and G2, P3 and G3.

[0060] The method (500) then includes optimizing (503) a

disaggregation ratio to prepare the forecast at a forecast level. In one example optimizing (503) a disaggregation ratio to prepare the forecast at a forecast level may include calculating a minimum error forecast. In this example, the minimum error forecast (mef) may be defined as follows:

mef = min [(actG4P4)-dfGnPn] (equation 1 ) where df is the disaggregation forecast from a geography dimension n level (Gn) and a production dimension m level (Pm). In this example, n is the level in a geography dimension, and m is the levels in a product dimension. In this example, four is the highest level for n and m. Further, in this example, df can be an estimate from actual historical values or six month moving averages. [0061] In one example, optimizing a disaggregation ratio to prepare the forecast at a forecast level includes using the minimum error forecast. In this example, a disaggregation ratio (odr) may be defined as follows:

odr = mef/ (∑ mefl +mef2+...+ mefn) (equation 2) where n is the lowest level of geography dimensions. As a result, a

disaggregation ratio is used to prepare the forecast. Further, in one example, several disaggregation ratios may be used to prepare the forecast. In this example, the method (500) may optimized a disaggregation ratio such that one disaggregation ratio is used to prepare the forecast at a forecast level.

[0062] As mentioned above, a forecast level may be a level at which a forecast is prepared. In one example, a forecast level may a SKU level. As a result, a forecast is prepared at a SKU level. In another example, a forecast level may a category level. As a result, a forecast is prepared at a category level. In yet another example, a forecast level may an aggregate level. As a result, a forecast is prepared at an aggregate level. Further, in one example, a user may select the forecast level for preparing the forecast. As a result, a forecast may be prepared at a desired forecast level.

[0063] Fig. 6 is a flowchart of an example of a method for optimizing a forecast with a disaggregation ratio at a forecast level, according to one example of principles described herein. In this example, the method (600) includes determining (601 ) product dimensions and geography dimensions to be evaluated in a forecast, evaluating (602) combinations of the product dimensions and geography dimensions, disaggregating (603) at least one minimum error forecast for at least one disaggregation method based on the combinations of the product dimensions and the geography dimensions, identifying (604) an optimal disaggregation method, calculating (605) an optimal disaggregation path for the product dimensions and the geography dimensions based on the minimum error forecast, and optimizing (606) a disaggregation ratio to prepare the forecast at a forecast level.

[0064] As mentioned above, the method (600) includes

disaggregating (603) at least one minimum error forecast for at least one disaggregation method based on the combinations of the product dimensions and the geography dimensions. As mentioned above, in preparing a forecast, a forecaster decides at what level forecasts are to be used. For example, a forecast may be used at a SKU level, category level, or aggregate level. Further, forecasting done at the lowest forecast levels provides the best input to market demand and captures more information about the actual demand of the market, but the time and effort to carry out such an exercise makes this type of forecast impractical. Choosing a higher forecast level is a tradeoff between flexibility, time, and effort used for actual forecasting process. Further, the forecaster decides which disaggregation method is to be used in preparing the forecast. In one example, a disaggregation method may be a top-down disaggregation method, a middle-out disaggregation method, a bottom-up disaggregation method, or combinations thereof. As a result, different disaggregation methods may be used to prepare an accurate forecast.

[0065] In one example, a top-down disaggregation method for a forecast is prepared first at the highest forecast level of aggregation, such as a company as a whole and then disaggregated into categories and SKUs. For example, in a top-down disaggregation method an overview of the business is formulated, specifying but not detailing, any first-level subsystems of the business. Each subsystem is then refined in yet greater detail, sometimes in many additional subsystem levels, until the entire specification is reduced to base elements such as categories and SKUs.

[0066] In keeping with the given example, a middle-out

disaggregation method for a forecast is prepared first at a forecast level such as a category level. Further, the middle-out disaggregation method then develops a forecast at a company level in which forecasts of all the categories are added to the forecast. In a bottom-up disaggregation method for a forecast, the forecast is prepared first at a forecast level, such as a SKU forecast level, and then aggregated to generate a category forecast level and aggregate forecast level forecast.

[0067] As mentioned above, the method (600) further includes identifying (604) an optimal disaggregation method. In one example, choosing the right forecasting level to generate a forecast and methods to disaggregate can significantly affect the accuracy of the forecast with downstream

implications in supply chain and inventory management. With changing market conditions and shorter product cycles, excess inventory and lost sales opportunities can significantly hurt a business. As a result, identifying an optimal disaggregation method can significantly impact a forecast's accuracy.

[0068] In one example, the top-down disaggregation method may be the optimal disaggregation method. In another example, the middle-out disaggregation method may be the optimal disaggregation method. In yet another example the bottom-up disaggregation method may be the optimal disaggregation method.

[0069] The method further includes calculating (605) an optimal disaggregation path for the product dimensions and the geography dimensions based on the minimum error forecast. As mentioned above, an identification engine may be used to calculate an optimal disaggregation path for the product dimensions and the geography dimensions based on the minimum error forecast. In one example, an optimal disaggregation path is calculated for every combination of the product dimensions and the geography dimensions. In another example, an optimal disaggregation path is calculated for one product dimension and every combination of the geography dimensions. In yet another example, an optimal disaggregation path is calculated for every combination of the product dimensions and for one geography dimension.

[0070] Fig. 7 is a diagram of an example of an optimizing system, according to one example of principles described herein. The optimizing system (700) includes a determining engine (702), an evaluating engine (704), and an optimizing engine (706). In this example, the optimizing system (700) also includes a disaggregating engine (708), an identifying engine (710), and a calculating engine (712). The engines (702, 704, 706, 708, 710, 712) refer to a combination of hardware and program instructions to perform a designated function. Each of the engines (702, 704, 706, 708, 710, 712) may include a processor and memory. The program instructions are stored in the memory and cause the processor to execute the designated function of the engine. [0071] The determining engine (702) determines product dimensions and geography dimensions to be evaluated in a forecast. In one example, the determining engine (702) determines all product dimensions and all geography dimensions to be evaluated in a forecast.

[0072] The evaluating engine (704) evaluates combinations of the product dimensions and geography dimensions. In one example, the evaluating engine (704) evaluates all possible combinations of the product dimensions and geography dimensions.

[0073] The optimizing engine (706) optimizes a disaggregation ratio to prepare the forecast. In one example, the optimizing engine (706) optimizes one disaggregation ratio to prepare the forecast. In another example, the optimizing engine (706) optimizes several disaggregation ratios to prepare the forecast at a forecast level.

[0074] The disaggregating engine (708) disaggregates at least one minimum error forecast for at least one disaggregation method based on the combinations of the product dimensions and the geography dimensions. In one example, the disaggregation method includes a top-down disaggregation method, a middle-out disaggregation method, a bottom-up disaggregation method, or combinations thereof.

[0075] The identifying engine (710) identifies an optimal

disaggregation method. In one example, the optimal disaggregation method may be a top-down disaggregation method, a middle-out disaggregation method, a bottom-up disaggregation method, or combinations thereof.

[0076] The calculating engine (712) calculates an optimal

disaggregation path for the product dimensions and the geography dimensions based on the minimum error forecast. In one example, the calculating engine (712) calculates one optimal disaggregation path for the product dimensions and the geography dimensions based on the minimum error forecast. In another example, the calculating engine (712) calculates several optimal disaggregation paths for the product dimensions and the geography dimensions based on the minimum error forecast. [0077] Fig. 8 is a diagram of an example of an optimizing system, according to one example of principles described herein. In this example, optimizing system (800) includes processing resources (802) that are in communication with memory resources (804). Processing resources (802) include at least one processor and other resources used to process

programmed instructions. The memory resources (804) represent generally any memory capable of storing data such as programmed instructions or data structures used by the optimizing system (800). The programmed instructions shown stored in the memory resources (804) include a product dimension determiner (806), a geography dimension determiner (808), a combination evaluator (810), a minimum error forecast disaggregater (812), an optimal disaggregation method identifier (814), a top-down disaggregater (816), a middle-out disaggregater (818), a bottom-up disaggregater (820), an optimal disaggregation path calculator (822), and a disaggregation ratio optimizer (824).

[0078] The memory resources (804) include a computer readable storage medium that contains computer readable program code to cause tasks to be executed by the processing resources (802). The computer readable storage medium may be tangible and/or physical storage medium. The computer readable storage medium may be any appropriate storage medium that is not a transmission storage medium. A non-exhaustive list of computer readable storage medium types includes non-volatile memory, volatile memory, random access memory, write only memory, flash memory, electrically erasable program read only memory, or types of memory, or combinations thereof.

[0079] The product dimension determiner (806) represents

programmed instructions that, when executed, cause the processing resources (802) to determine a product dimension. The geography dimension determiner (808) represents programmed instructions that, when executed, cause the processing resources (802) to determine a geography dimension. The combination evaluator (810) represents programmed instructions that, when executed, cause the processing resources (802) to evaluate combinations of product dimensions and geography dimensions. The minimum error forecast disaggregater (812) represents programmed instructions that, when executed, cause the processing resources (802) to disaggregate a minimum error forecast. The optimal disaggregation method identifier (814) represents programmed instructions that, when executed, cause the processing resources (802) to identify an optimal disaggregation method.

[0080] The top-down disaggregater (816) represents programmed instructions that, when executed, cause the processing resources (802) to disaggregate using a top-down disaggregation method. The middle-out disaggregater (818) represents programmed instructions that, when executed, cause the processing resources (802) to disaggregate using a middle-out disaggregation method. The bottom-up disaggregater (820) represents programmed instructions that, when executed, cause the processing resources (802) to disaggregate using a bottom-up disaggregation method. The optimal disaggregation path calculator (822) represents programmed instructions that, when executed, cause the processing resources (802) to calculate an optimal disaggregation path. The disaggregation ratio optimizer (824) represents programmed instructions that, when executed, cause the processing resources (802) to optimize a disaggregation ratio.

[0081] Further, the memory resources (804) may be part of an installation package. In response to installing the installation package, the programmed instructions of the memory resources (804) may be downloaded from the installation package's source, such as a portable medium, a server, a remote network location, another location, or combinations thereof. Portable memory media that are compatible with the principles described herein include DVDs, CDs, flash memory, portable disks, magnetic disks, optical disks, other forms of portable memory, or combinations thereof. In other examples, the program instructions are already installed. Here, the memory resources can include integrated memory such as a hard drive, a solid state hard drive, or the like.

[0082] In some examples, the processing resources (802) and the memory resources (804) are located within the same physical component, such as a server, or a network component. The memory resources (804) may be part of the physical component's main memory, caches, registers, non-volatile memory, or elsewhere in the physical component's memory hierarchy.

Alternatively, the memory resources (804) may be in communication with the processing resources (802) over a network. Further, the data structures, such as the libraries, may be accessed from a remote location over a network connection while the programmed instructions are located locally. Thus, the optimizing system (800) may be implemented on a user device, on a server, on a collection of servers, or combinations thereof.

[0083] The optimizing system (800) of Fig. 8 may be part of a general purpose computer. However, in alternative examples, the optimizing system (800) is part of an application specific integrated circuit.

[0084] The preceding description has been presented to illustrate and describe examples of the principles described. This description is not intended to be exhaustive or to limit these principles to any precise form disclosed. Many modifications and variations are possible in light of the above teaching.

Claims

CLAIMS WHAT IS CLAIMED IS:

1 . A method for optimizing a forecast with a disaggregation ratio at a forecast level, said method comprising:

determining product dimensions and geography dimensions to be evaluated in a forecast;

evaluating combinations of said product dimensions and said geography dimensions; and

optimizing a disaggregation ratio to prepare said forecast at a forecast level.

2. The method of claim 1 , further comprising disaggregating at least one minimum error forecast for at least one disaggregation method based on said combinations of said product dimensions and said geography dimensions.

3. The method of claim 2, wherein said at least one said

disaggregation method includes a top-down disaggregation method, a middle-out disaggregation method, a bottom-up disaggregation method, or combinations thereof.

4. The method of claim 2, wherein disaggregating said at least one said minimum error forecast for said at least one said disaggregation method based on said combinations of said product dimensions and said geography dimensions further comprising identifying an optimal disaggregation method.

5. The method of claim 2, wherein disaggregating said at least one said minimum error forecast for said at least one said

disaggregation method based on said combinations of said product dimensions and said geography dimensions further comprising calculating an optimal disaggregation path for said product dimensions and said geography dimensions based on said minimum error forecast.

6. The method of claim 1 , wherein said geography dimensions

include a region, a market segment, a sub region, a country, or combinations thereof.

7. The method of claim 6, wherein said geography dimensions

includes historical values for said region, historical values for said market segment, historical values for said sub region, historical values for said country, or combinations thereof.

8. A system for optimizing a forecast with a disaggregation ratio at a forecast level, said system comprising:

a determining engine to determine product dimensions and geography dimensions to be evaluated in a forecast;

an evaluating engine to evaluate combinations of said product dimensions and said geography dimensions;

an identifying engine to identify an optimal disaggregation method; and

an optimizing engine to optimize a disaggregation ratio to prepare said forecast at a forecast level.

9. The system of claim 8, further comprising a disaggregating engine to disaggregate at least one minimum error forecast for at least one disaggregation method based on said combinations of said product dimensions and said geography dimensions.

10. The system of claim 9, wherein said at least one said

1 1 . The system of claim 8, further comprising a calculating engine to calculate an optimal disaggregation path for said product dimensions and said geography dimensions based on said minimum error forecast.

12. The system of claim 8, wherein said geography dimensions

includes a region, a market segment, a sub region, a country, or combinations thereof and wherein said geography dimensions further include historical values for said region, historical values for said market segment, historical values for said sub region, historical values for said country, or combinations thereof.

13. A computer program product for optimizing a forecast with a

disaggregation ratio at a forecast level, comprising:

a tangible computer readable storage medium, said tangible computer readable storage medium comprising computer readable program code embodied therewith, said computer readable program code comprising program instructions that, when executed, causes a processor to:

determine product dimensions and geography dimensions to be evaluated in a forecast;

evaluate combinations of said product dimensions and said geography dimensions; disaggregate at least one minimum error forecast for at least one disaggregation method based on said combinations of said product dimensions and said geography dimensions; and optimize a disaggregation ratio to prepare said forecast at a forecast level.

14. The product of claim 13, further comprising computer readable program code comprising program instructions that, when executed, cause said processor to disaggregate at least one minimum error forecast for at least one disaggregation method based on said combinations of said product dimensions and said geography dimensions.

15. The product of claim 13, further comprising computer readable program code comprising program instructions that, when executed, cause said processor to identify an optimal

disaggregation method wherein said optimal disaggregation method includes a top-down disaggregation method, a middle-out disaggregation method, a bottom-up disaggregation method, or combinations thereof.