US20140081707A1

US20140081707A1 - Disambiguating point-of-sale data through item indexing

Info

Publication number: US20140081707A1
Application number: US14/085,800
Authority: US
Inventors: Robert J. Pressini; William D. Dunlavy; Ronald S. Sorenson
Original assignee: Stafanpolus KG LLC
Current assignee: Intellectual Ventures Assets 186 LLC
Priority date: 2003-05-09
Filing date: 2013-11-21
Publication date: 2014-03-20
Also published as: US7292991B2; US8630909B2; US20070124195A1; US20040225568A1; WO2004102342A2; WO2004102342A3

Abstract

Methods, apparatus, and computer-readable medium can disambiguate point-of-sale data. One method receives one or more measured values. Each measured value indicating a quantity of sales allocated to a respective individual item. The method also receives an unallocated value indicating a quantity of sales allocated to an undifferentiated grouping of a plurality of the individual items. The method calculates, for each individual item of the plurality of the individual items associated with the undifferentiated grouping, a fraction of the undifferentiated grouping attributable to the respective individual item, and estimates a total sales value for at least one of the plurality of the individual items, based at least in part upon a respective measured value and a respective fraction of the undifferentiated grouping attributable to the respective individual item.

Description

CROSS-REFERENCE TO RELATED APPLICATION

This application is a continuation of, and claims priority under 35 U.S.C. §120 to application Ser. No. 11/616,700 entitled “DISAMBIGUATING POINT-OF-SALE DATA THROUGH ITEM INDEXING”, filed on Dec. 27, 2006, now issued as U.S. Pat. No. ______, which is a continuation of, and claims the priority benefit, under 35 U.S.C. §120, of U.S. application Ser. No. 10/434,795, entitled “DISAMBIGUATING POINT-OF-SALE DATA THROUGH ITEM INDEXING” filed on May 9, 2003, now issued as U.S. Pat. No. 7,292,991. The subject matter of these earlier filed applications are hereby incorporated by reference.

TECHNICAL FIELD

The present invention relates generally to the fields of demand forecasting and production planning. More specifically, the present invention relates to techniques for increasing forecast accuracy by disambiguating point-of-sale data.

BACKGROUND OF THE INVENTION

Demand forecasting is an essential component of production planning. Nowhere is this statement more true than in the perishable food industry. Bakeries, delis, doughnut shops, supermarkets, and the like, must accurately predict customer demand in order to produce the right amount of goods to be sold each day. Forecasts that are too high typically result in wasted goods. Forecasts that are too low often result in lost sales and customer annoyance.
Various software programs, such as Fresh Market Manager™ (FMM), available from Park City Group, Inc., have addressed the need for demand forecasting. The FMM forecasting engine predicts how many of a given item is required each day based on historical point-of-sale data, adjusted for influences such as holidays, weather, seasons, and competitor activity.
Unfortunately, the forecasts produced by these software programs are only as accurate as the data provided to them. One difficulty in obtaining accurate data is due to the fact that the sale of certain items can be recorded in different ways by a point-of-sale device (e.g., cash register). For example, certain individual items, such as chocolate chip cookies, can be “rung up” as miscellaneous or variety items, e.g., “cookies” or “miscellaneous bakery.” However, the forecasting engine does not need to forecast how many miscellaneous bakery items are needed in a given day. Rather, it must determine how many chocolate chip cookies will be required to satisfy customer demand.
Point-of-sale data that include both individual item sales and undifferentiated, miscellaneous or variety item sales are ambiguous and typically result in inaccurate forecasts. Accordingly, a need exists for a technique for disambiguating point-of-sale data to enhance forecasting accuracy.

SUMMARY OF THE INVENTION

A point-of-sale interface receives sales quantities for a number of individual items from a point-of-sale feed. The point-of-sale interface also receives sales quantities for a miscellaneous group item that includes undifferentiated sales of one or more of the individual items. Optionally, the point-of-sale interface excludes sales marked as special orders from the sales quantities.
An item indexer calculates an item index for each individual item comprising a fraction of the miscellaneous group item sales attributable to the individual item. In one embodiment, the item index is calculated by determining a quantity of the individual item produced but not accounted for in the sales for the individual item and any non-point-of-sale transfers (e.g., invoice orders, shrink/waste), and dividing the quantity of the individual item produced but not accounted for by a sum of all of the individual items produced but not accounted for in the sales for the individual items and any non-point-of-sale transfers.
A sales disambiguator computes a revised sales quantity for each individual item using the item index and the sales for the miscellaneous group item and the individual item. In one embodiment, the revised sales quantity is calculated by adding the original sales quantity for the individual item to a product of the item index for the individual item and the point-of-sale quantity for the miscellaneous group item.
The sales disambiguator provides the revised sales quantities for the Individual items to a forecasting system, resulting in a more accurate forecast than can be had by relying on the original sales quantities.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block diagram of a conventional system for forecasting and production planning;

FIG. 2 is a block diagram of a point-of-sale (POS) feed;

FIG. 3 is a block diagram of a forecasting and production planning system according to an embodiment of the invention;

FIG. 4 is a block diagram of a POS interface;

FIG. 5 is a block diagram of an item indexer.

FIG. 6 is a dataflow diagram showing the creation of item indices;

FIG. 7 is a dataflow diagram showing the creation of disambiguated sales quantities based on the item indices;

FIG. 8 is a table comparing disambiguated sales quantities with estimated quantities using a conventional approach; and

FIG. 9 is a flowchart of a method for disambiguating point-of-sale data to increase forecasting accuracy.

DETAILED DESCRIPTION

Reference is now made to the figures in which like reference numerals refer to like elements. For clarity, the first digit of a reference numeral indicates the figure number in which the corresponding element is first used.
In the following description, numerous specific details of programming, software modules, user selections, network transactions, database queries, database structures, etc., are provided for a thorough understanding of the embodiments of the invention. However, those skilled in the art will recognize that the invention can be practiced without one or more of the specific details, or with other methods, components, materials, etc.
In some cases, well-known structures, materials, or operations are not shown or described in detail in order to avoid obscuring aspects of the invention. Furthermore, the described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
In a conventional forecasting and production planning system 100, illustrated in FIG. 1, a forecasting engine 102 receives historical sales data 104 directly from a point-of-sale (POS) feed 106. The POS feed 106 may be implemented on a network server, which receives daily sales or transaction-level sales for one or more products from a number of POS devices 108 (e.g., cash registers). The POS feed 106 may be accessible, for example, through a local area network (LAN) or the Internet.
The forecasting engine 102 uses the historical sales data 104 to generate a forecast 110 of consumer demand for each product. The forecasting engine 102 may refine the forecast 110 using various external data 112, such as information regarding holidays, weather, seasons, competitor activity, etc.
A production planning engine 114 uses the forecast 110 to create a plan 116 for producing a sufficient quantity of each product to satisfy the forecasted demand. Depending on the industry, the production plan 116 may include, for example, a list of required ingredients, a labor schedule, etc.
A production facility 118, such as a bakery or deli, uses the production plan 116 to produce the forecasted quantities of products, which are then inventoried and sold in due course. A cash register or other POS device 108 records the daily number of sales of each product and periodically sends the sales quantities to the POS feed 106 for future access by the forecasting engine 102.
Referring to FIG. 2, a typical POS feed 106 provides daily or transaction-level sales 202 for a number of “individual” items, i.e., items whose sale can be specifically recorded and for which a specific forecast of demand is needed. An example of an individual item is a chocolate chip cookie, since a POS device 106 may have a designated code or button for recording the sale, and a bakery needs to know how many chocolate chip cookies to produce each day. In some cases, an individual item may include a group of other individual items sold as a unit, e.g., a cookie tin, since a forecast 110 of the number of cookie tins to produce in a given day may be desirable.
The POS feed 106 may also provide daily or transaction-level sales 204 for one or more “miscellaneous group” items. Many POS devices 106 allow sales of Individual items to be recorded under “miscellaneous” or “variety” designations. This can be helpful to a check-out clerk who does not know the specific code for an individual item. For example, a chocolate chip cookie may be rung up as a “cookie” or as a “miscellaneous bakery” item. This is also required when a group of items can only be sold using a single code. For example, varieties of bagels (blueberry bagel, onion bagel, plain bagel, etc.) may only be sold using a ‘Bagel’ code.
However, the existence of miscellaneous group items on a cash register is problematic to forecasting and production planning. Because sales of miscellaneous group items comprise undifferentiated sales of one or more of the individual items, the resulting historical sales data 104 are ambiguous. For example, as shown in FIG. 2, sixty cookies were rung up as a generic “cookie” item, as opposed to the particular variety of cookie, which represents almost two-thirds of the reported sales.
A conventional forecasting engine 102 might estimate the miscellaneous cookie sales attributable to the individual varieties based on the relative proportions of the individual item sales quantities 202. However, such an estimate can be very inaccurate. For example, suppose that, of the sixty generic cookie sales, fifty were oatmeal cookies. A forecast 110 based on a proportional allocation of the miscellaneous group item sales would be inaccurately weighted to the other varieties.
FIG. 3 is a block diagram of a forecasting and production planning system 300 according to an embodiment of the invention that solves the aforementioned problems and disadvantages. A point-of-sale (POS) interface 302 receives the historical sales data 104 from the POS feed 106. The historical sales data 104 includes individual and miscellaneous group item sales quantities 202, 204 for the same particular time period, such as a day or a portion of a day.
As described below, the POS interface 302 may optionally include logic for scrubbing, removing, or otherwise blocking certain types of data. For example, special orders are likely to be one-time events. Hence, the POS interface 302 may automatically exclude sales quantities 202, 204 marked as special orders, generating scrubbed sales data 304.
An item indexer 306 receives the scrubbed sales data 304 and creates an item index 308 for each individual item. As explained in greater detail hereafter, an item index 308 is a fraction or percentage contribution of an individual item to the sales of an associated miscellaneous group item. In one embodiment, the item indexer 306 relies on production data 310 from the production facility 118 to create the indices 308.
A sales disambiguator 312 uses the item indices 308 and the scrubbed sales data 304 to generate disambiguated or revised sales quantities 314, representing the actual quantity of each product sold. The disambiguated sales quantities 314 may then be fed to the forecasting engine 102 to provide a more accurate forecast 316 of demand to the production planning engine 114.
Both the forecasting and production planning engines 102, 114 may be implemented using components of an available software package, such as Fresh Market Manager™ (FMM), sold by Park City Group, Inc., of Park City, Utah. However, the use of other forecasting and production planning systems are contemplated within the scope of the invention. Furthermore, any of the modules, engines, components, etc., described herein may be implemented using a suitable combination of hardware, software, and/or firmware known to those of ordinary skill in the art.
FIG. 4 illustrates an optional data-scrubbing function of the POS interface 302. As noted, historical sales data 104 may include sales quantities 202, 204 for individual and miscellaneous group items. In one embodiment, certain sales quantities 202, 204 may be tagged or marked as a “special order,” i.e., an order placed and/or fulfilled outside of the usual routine, possibly for customized products, that is not likely to recur with any degree of predictability. An example of a special order at a bakery may be three-dozen red sugar cookies to celebrate the Superbowl victory of Tampa Bay. Any suitable marking or tagging method may be used within the scope of the invention, such as an eXtensible Markup Language (XML) tag or the like.
Special orders are likely to be one-time events, which makes them of little value from a forecasting perspective. Accordingly, the POS Interface 302 identifies sale quantities 202, 204 marked as special orders. The marked sales quantities 202, 204 are then scrubbed, removed, masked, or otherwise excluded by the POS interface 302. The resulting scrubbed sales data 304 may then be provided to the item indexer 306 and sales disambiguator 312 as described above.
Of course, the data-scrubbing function of the POS interface 302 may be configured in other ways and/or disabled by an operator. Moreover, in alternative embodiments, the item indexer 306 and/or the sales disambiguator 312 may perform the data-scrubbing function.
FIG. 5 illustrates further details of item indexer 306. As noted above, the item indexer 306 receives the scrubbed sales data 304 and the production data 310. In one embodiment, the production data 310 includes, for each individual item, a production quantity 502, a shrink/waste quantity 504, and an invoice order quantity 506. The production quantity 502 is the quantity of each item produced, while the shrink/waste quantity 504 represents the quantity of each item that was discarded either prior to or after packaging.
The invoice order quantity 506 corresponds to the quantity of each item that was sold by an invoice order, as opposed to a retail, over-the-counter sale. An invoice order differs from a special order in that an invoice order typically occurs with predicable frequency. Moreover, an invoice order usually pertains to standard products, e.g., three-dozen blueberry bagels.
The item indexer 306 uses the production data 310, including the production quantities 502, shrink/waste quantities 504, and/or invoice order quantities 506, to generate an item index 308 for each individual item. In the depicted embodiment, an item index 308 is a number between zero and one, and represents fraction or percentage contribution of an individual item to the quantity sales 204 of an associated miscellaneous group item (e.g., “cookies”).
One algorithm for calculating the item indices is 308 set forth in the following pseudocode. Those of skill in the art will recognize that the algorithm may be implemented using any conventional programming language, such as C++. Java, or the like.
For Each Miscellaneous Group Item:


Sum = 0
For each Individual Item:
Misc. Net Quantity of Individual Item (MNQ) = Production Total −
Shrink/Waste Total − Invoice Orders Total − Individual Item Sales
Total
Sum = Sum + MNQ
End for (Individual Item)
For each individual Item:
Index = MNQ / Sum
End for (Individual Item)
End for (Miscellaneous Group Item)

FIG. 6 illustrates the above-identified algorithm in connection with sample data. For simplicity, only a single miscellaneous group item is used in this example. However, those of skill in the art will recognize that the techniques disclosed herein may be used for multiple miscellaneous group items.
Initially, the item indexer 306 subtracts the shrink/waste quantities 504, invoice order quantities 506, and individual item sales quantities 202 from the production quantities 502 for each individual item. The resulting figure is a “miscellaneous net quantity” (MNQ) 602 of each individual item that was not accounted for in the individual item sales quantities 202 or by non-point-of-sale transfers (e.g., invoice orders, shrink/waste).
While determining the MNQ 602 for each individual item, a running sum 604 of the MNQs for all of the individual items is made. The item index 308 for a particular individual item is then calculated by dividing the MNQ 602 for that item by the MNQ sum 604 for all of the individual items.
In the depicted example, 100 chocolate chip cookies were produced, of which 20 were discarded, none were sold by invoice order, and 30 were sold at the cash register. Hence, 50 chocolate chip cookies are not accounted for (the MNQ 602). The sum of the MNQs 602 for all cookie varieties is 100. Accordingly, the item index 308 for the chocolate chip cookies, as a percentage or fraction of the sales quantities 204 for the “cookies” miscellaneous group item is 0.5.
Referring to FIG. 7, the sales disambiguator 312 uses the item indices 308 to calculate disambiguated sales quantities 314 for each of the individual items. With respect to each individual item, the sales disambiguator 312 adds the sales quantity 202 for the individual item to a product of the item index 308 for the individual item and the sales quantity 204 for the associated miscellaneous group item. After all of the disambiguated sales quantities 314 are calculated, they are provided to the forecasting engine 102.
In the illustrated example, the item index 308 for chocolate chip cookies is 0.5. In addition, the sales quantity 204 for the “cookies” miscellaneous group item is 60, and the sales quantity 204 for chocolate chip cookies, as recorded at the point-of-sale, is 30. Based on these data, the disambiguated sales quantity 314 for chocolate chip cookies is 60.
FIG. 8 provides a comparison of the disambiguated sales quantities 314 of FIG. 7 with estimated sales quantities based on a proportionate allocation of the miscellaneous group item sales quantities 204 among the sales quantities 202 for the various individual items. As may be observed, the estimated sales quantity for oatmeal cookies is 27% higher than the actual, disambiguated sales quantity 314. Even more significant is the estimated sales quantity for sugar cookies, which is 42% less than the actual, disambiguated sales quantity 314.
In view of the foregoing, a forecast 110 based on the estimated sales quantities, as in conventional approaches, would be highly inaccurate, resulting in the production of too many of some varieties (e.g., peanut butter cookies and sugar cookies), while not producing enough of other varieties (e.g., chocolate chip cookies and oatmeal cookies).
FIG. 9 is a flowchart of a method 900 for disambiguating point-of-sale data that summarizes techniques described above. For simplicity, as in FIGS. 6 and 7, only one miscellaneous group item is referenced, although the invention is not limited in this respect.
Initially, the system 300 receives 902 historical point-of- sale sales 202, 204 from the POS feed 106. Thereafter, for each miscellaneous group item 904, and then for each individual item 906, the system 300 calculates 908 an item index 308 for the individual item by dividing a quantity of the individual item produced but not accounted for in point-of-sale quantities for the individual item and any non-point-of-sale transfers (e.g., invoice orders, shrink/waste) by the sum of all individual items produced but not accounted for.
Next, the system 300 computes 910 a disambiguated sales quantity 314 for the individual item (representing a quantity sold) by adding the sales quantity 202 for the individual item to a product of the item index for the individual item and the sales quantity 204 for the miscellaneous group item. Once all of the disambiguated sales quantities 314 have been calculated, the system 300 provides 912 the disambiguated sales quantities 314 to the forecasting engine 102.
While specific embodiments and applications of the present invention have been illustrated and described, it is to be understood that the invention is not limited to the precise configuration and components disclosed herein. Various modifications, changes, and variations apparent to those of skill in the art may be made in the arrangement, operation, and details of the methods and systems of the present invention disclosed herein without departing from the spirit and scope of the present invention.

Claims

What is claimed is:

1. A method comprising:

receiving one or more measured values, each measured value indicating a quantity of sales allocated to a respective individual item;

receiving an unallocated value indicating a quantity of sales allocated to an undifferentiated grouping of a plurality of the individual items;

calculating, for each individual item of the plurality of the individual items associated with the undifferentiated grouping, a fraction of the undifferentiated grouping attributable to the respective individual item; and

estimating, for at least one of the plurality of the individual items, a total sales value based at least in part upon a respective measured value and a respective fraction of the undifferentiated grouping attributable to the respective individual item.

2. The method of claim 1, further comprising:

providing the total sales value for the at least one of the plurality of the individual items to a sales forecasting system.

3. The method of claim 1, further comprising:

scrubbing the one or more measured value to remove a portion of the quantity of sales allocated to the respective individual item that are deemed to potentially distort the respective estimate of total sales value.

4. The method of claim 3, wherein scrubbing comprises removing a portion of the quantity of sales that are associated with a special order of the respective individual item.

5. The method of claim 1, wherein calculating a fraction of the undifferentiated grouping attributable to the respective individual item comprises:

determining a production value for each individual item;

determining an un-sold value for each individual item;

calculating, for each respective individual item, an item's unallocated value by subtracting from the production value, associated with the respective individual item, at least the un-sold value, associated with the respective individual item, and the measured value, associated with the respective individual item; and

calculating, for each respective individual item, the fraction of the undifferentiated grouping attributable to the respective individual item by dividing the unallocated value, associated with the respective individual item, by a sum of all the unallocated values of the individual items.

6. The method of claim 5, wherein estimating, for at least one of the plurality of the individual items, a total sales value further comprises:

determining a transferred value for each of the at least one individual item, wherein the transferred value indicates a quantity that is not included in the respective measured value; and

wherein calculating, for each respective individual item, an item's unallocated value also comprises subtracting the transferred value associated with the respective individual item from the production value, associated with the respective individual item.

7. The method of claim 1, wherein calculating a fraction of the undifferentiated grouping attributable to the respective individual item comprises:

determining the fraction of the undifferentiated grouping attributable to the respective individual item based, at least in part, upon a quantity of each individual item produced and an amount of each individual item directly accounted for.

8. The method of claim 1, wherein estimating, for at least one of the plurality of the individual items, a total sales value comprises:

computing an approximately-allocated value for each of the at least one individual item, by allocating to each of the individual items a portion of the unallocated value based upon the respective fraction of the undifferentiated grouping attributable to the respective individual item; and

adding, for each of the at least one individual item, the respective approximately-allocated value to a respective measured value to create the total sales value associated with the respective individual item.

9. The method of claim 1, wherein both the measured value and the unallocated value for each respective individual item are associated with a same period of time.

10. The method of claim 1, wherein at least one measured value is associated with a grouped item, wherein the grouped item comprises a plurality of other individual items; and

wherein receiving one or more measured values comprises allocating the measured value associated with the grouped item to two or more measured values associated with the other individual items included by the grouped item.

11. A computer program product for estimating sales information, the computer program product being tangibly and non-transitorily embodied on a computer-readable medium and including executable code for execution on a data processing apparatus, the executable code comprising:

instructions to receive one or more measured values, each measured value indicating a quantity of sales allocated to a respective individual item;

instructions to receive an unallocated value indicating a quantity of sales allocated to an undifferentiated grouping of a plurality of the individual items;

instructions to calculate, for each individual item of the plurality of the individual items associated with the undifferentiated grouping, a fraction of the undifferentiated grouping attributable to the respective individual item; and

instructions to estimate, for at least one of the plurality of the individual items, a total sales value, based at least in part upon a respective measured value and a respective fraction of the undifferentiated grouping attributable to the respective individual item.

12. The computer program product of claim 11, wherein the executable code further comprises:

instructions to cause the data processing apparatus to provide the total sales values for each of the plurality of the individual items to a sales forecasting system.

13. The computer program product of claim 11, wherein the executable code further comprises:

instructions to cause the data processing apparatus to scrub the one or more measured value to remove a portion of the quantity of sales allocated to respective individual item that are deemed to potentially distort the respective estimate of total sales value.

14. The computer program product of claim 11, wherein the executable code further comprises:

instructions to cause the data processing apparatus to calculate a fraction of the undifferentiated grouping attributable to the respective individual item by determining the fraction of the undifferentiated grouping attributable to the respective individual item based, at least in part, upon a quantity of each individual item produced and an amount of each individual item directly accounted for.

15. The computer program product of claim 1, wherein the executable code further comprising:

instructions to cause the data processing apparatus to estimate a total sales values for each of the plurality of the individual items by:

16. An apparatus comprising:

an item indexer configured to:

receive one or more measured values, each measured value indicating a quantity of sales allocated to a respective individual item,

receive an unallocated value indicating a quantity of sales allocated to an undifferentiated grouping of a plurality of the individual items, and

calculate, for each individual item of the plurality of the individual items associated with the undifferentiated grouping, a fraction of the undifferentiated grouping attributable to the respective individual item; and

a sales disambiguator configured to estimate, for at least one of the plurality of the individual items, a total sales value based at least in part upon a respective measured value and a respective fraction of the undifferentiated grouping attributable to the respective individual item.

17. The apparatus of claim 16, further comprising:

a production forecaster configured to predict consumer demand for each individual item based, at least in part, upon the total sales value for each of at least one of the plurality of the individual items estimated by the sales disambiguator.

18. The apparatus of claim 16, further comprising:

a data scrubber configured to scrub the one or more measured values to remove a portion of the quantity of sales allocated to respective individual item that are deemed to potentially distort the respective estimate of total sales value.

19. The apparatus of claim 16, wherein the item indexer is configured to calculate a fraction of the undifferentiated grouping attributable to the respective individual item based, at least in part, upon a quantity of each individual item produced and an amount of each individual item directly accounted for.

20. The apparatus of claim 16, wherein the sales disambiguator is configured to estimate, for at least one of the plurality of the individual items, a total sales value by:

computing an approximately-allocated value for each of the at least one of the plurality of the individual items, by allocating to each of the at least one of the plurality of the individual items a portion of the unallocated value based upon the respective fraction of the undifferentiated grouping attributable to the respective individual item; and

adding, for the at least one of the plurality of the individual item, the respective approximately-allocated value to a respective measured value to create the total sales value associated with the respective individual item.