Product Demand System and Method

This invention relates to an improved system for the analysis and prediction of product demand.

For more efficient product planning in a retail environment given a plurality of products of short shelf life, a tendency exists toward resorting to production plans that are calculated on an infrequent basis. This basis is often daily at best and more often is based on an average for a given day of the week. This results in excessive product and waste or not enough product to meet customer demand. To expect a human being to calculate product demand on a more frequent basis such as hourly or every fifteen minutes would be impractical if not impossible.

In addition to the impracticality of computing product demand manually on a more frequent basis is the complexity introduced by different demand patterns for each day of the week and seasonality during the year. Customer foot traffic and product preferences are unique to each day of the week. These traffic patterns and preferences are further complicated by seasonality trends week by week throughout the year. Accommodating this level of complexity requires a unique solution to store the past-product demand by seasonality, day of the week, and time period during the day, for each of a plurality of products.

Another difficulty encountered once the past-product demand has been stored is the ability to deal with incomplete data from current time periods and to compare trends in just-completed time periods against preestimated demand in the same time periods in order to adjust near-future time periods accurately and with confidence. One method for making such comparisons and projections on other near-future periods is to take a simple positive or negative percentage of the trend of just-completed periods against preestimated periods and apply it to the remaining preestimated near future periods for the day. This method is unreliable given the many anomalies that can occur, for example, when a bus load of people arrives or an unusually large order is placed.

It is therefore an object of the present invention to obviate the above mentioned drawbacks of the prior-art systems and to provide a system and method for analyzing product demand which incorporates tracking of past-product demand for a plurality of products, time intervals during the day, and other factors, and which introduces a confidence factor into product trends. Accordingly, more appropriate and accurate pre-estimation of future-product demand can be made.

FIG. 1 is a block schematic diagram of a workstation for estimating production in accordance with the present invention;

FIG. 2 is a data flow diagram showing the data used and stored during the update of past-product demand;

FIG. 3 is a flow chart of the process steps taken to post recent actual demand to the past-product demand curve;

FIG. 4 is a data flow diagram showing the data used and stored during the pre-estimating of product demand for near-future periods;

FIG. 5(comprised of 5A and 5B) is a flow chart of the process steps taken to pre-estimate product demand;

FIG. 6(comprised of 6A and 6B) is a flow chart of the process steps taken to monitor and revise near-future period product demand based on recent past periods of actual demand compared with the pre-estimate of the same recent past periods; and

FIG. 7 is a time chart illustrating operation of the present invention.

Referring now to FIG. 1, there is shown a block schematic diagram of a workstation configured in accordance with one embodiment of the present invention to provide improved production scheduling for a plurality of products. Central processing unit (CPU) 11 is connected to control the display device 13 in response to inputs supplied to the CPU 11 via the user's manipulation of the keyboard 15, or via other inputs supplied through a conventional network interface 29 (e.g., modem, bar-code readers, standard network bus, and the like). The CPU 11 accesses memory 18 which contains information that is supplied via the keyboard 15 or the network 29 (e.g., RS232 bus connection to a point-of-sale terminal), or is transferred thereto in conventional manner from mass storage medium 27. The memory contains instructions 19 for executing routines on information that is stored in several files 17 according to the present invention. The memory also includes main menu functions 23 and buffer storage or interactive display control 25 for synchronously displaying lines of data bits on display device 13 in conventional manner. Permanent records and periodic reports of product data may be plotted or printed out 31 under control of the CPU 11.

In accordance with the present invention, several files of information in data base 17 are gathered and stored regarding past demands for each of a plurality of products (e.g., cookies, buns, rolls, etc.) in each of a plurality of time periods (e.g., quarter-hour intervals per day). In addition, the data base 17 includes a product-planning master file, product planning frequency tables, rules calculation file, look-up table, a master file of demand-curve models, a master file of day types (e.g., Thanksgiving, Saturday, Superbowl day, etc.), a sales day file, a sales time file, a daily business summary file, an inventory details file, a cross-reference file, product planning work file, and location configuration file (e.g., number and types of ovens, mixers, counter area, etc.), all as more specifically described later herein. Files in the database 17 are accessed and modified under control of the CPU 11 in response to production planning routines 19 such as past demand update, product-demand pre-estimation, actual demand monitoring, and the like, as described later herein, for controlling the display on display device 13 of directions to a user for preparing specified quantities of individual products at prescribed time periods throughout the day.

Referring now to FIG. 2, there is shown a block schematic diagram of a system that operates to update past-product demand data. The process 10 of updating the product demand is described in FIG. 3. The demand-curve master file 14 defines each demand curve that will be tracked within each day type. Demand curves can represent total sales, department sales, individual product item sales, or counts of any kind. For example, as many as 999 individual demand curves can be defined. The past-product demand curve 16 is a file that is used to project near-future demand for products. The inventory detail transaction file 20 contains past inventory transactions such as receivers, transfers, waste, and on-hand. This detail can be used to update the past-product demand curve 16. The daily business summary file 22 contains daily historical actual product sales or production detail which also can be used to update the past product demand curve 16. The cross-reference file 24 operates between the products demand-curve master file 14 and the daily product business summary file 22, and allows for a single product demand curve to be updated by more than one daily product business summary. The sales-time file 26 contains past actual data by time period during the day for total sales, department sales, individual product sales, or counts of any kind. The sales-day file 28 contains the day-type indicator of the recent past actual data to determine which past-product demand curve 16 to update. The day-type master file 30 defines the day-type models (e.g., Saturdays, Thanksgiving, Superbowl day, and the like) that are available to track in the system of the present invention. The file descriptions are set forth in the attached Appendix.

Referring now to FIG. 3, there is shown a flow chart of the past-product demand routine. An unposted past demand day is determined by reading 32 the sales-day file 28 and finding an unposted day. The day-type model is retrieved 33 from the past-product demand curve 16 and the day-type master record is retrieved from the day-type master file 30. Next, the first demand curve product, or item, is retrieved 35 from the demand curve master file 14. The demand-curve master file 14 contains information regarding which recent-past actual data should be used and how it should be applied to the past-product demand curve 16. Recent-past actual data that can be accessed includes sales time file, daily product business summary, and inventory detail. Recent-past actual data is retrieved 37 from the sales time file 26 for developing a current-demand curve. The past-product demand curve 16 is retrieved 39, and the recent-past actual data is applied 41 to the past-product demand curve in accordance with the weighed percentage from the demand-curve master file 14. The updated past-product demand curve is re-written 43 to the past product demand file (SLSCUR) 16, and the day is marked 45 as posted (SLSDAY). For each demand-curve product or item in the master file 14, steps 35 through 43 are repeated. For each unposted past demand day, steps 32 through 45 are repeated.

FIGS. 4 and 5 illustrate the data stored and the manipulation of that data to pre-estimate product demand for near-future periods from past-product demand. FIG. 4 illustrates the flow of data in the process of pre-estimating the product demand. The process 38 of pre-estimating the product demand is described with reference to FIG. 5. The past-product demand curve file 14 and the sales day file 28 supply data for the process 38. The sales-day file 28 contains the day-type indicator of the near-future day to determine which past-product demand curve to use when pre-estimating product demand. The product planning master file 40 defines each individual product that must be planned and information as to how and when it should be planned. It also indicates which past-product demand curve is used with this planning estimation for this product. The product-planning frequency table #1 42 defines the frequency and what time of day a product should be produced. The product-planning frequency table #2 44 defines in more detail the frequency and time of day a product should be produced.

The process 46 of applying the rules to the past-product demand curve and daily sales projections to determine the pre-estimated product demand is described at step 57 in FIG. 5. The rule calculations file 48 contains information specific to the location and can be accessed by the calculation logic in the rule calculations file 48. The look-up table file 50 contains user-defined set of values or standards in table format that can be accessed by the calculation logic in the rule calculations file 48. Such user-defined values or standards may include minimum on-hand inventory counts, and minimum percentages of total sales to be of specified products, and the like. The product planning work file 51 is used to store pre-estimated product demand and revised product demand, as shown in the file descriptions in the attached Appendix.

Referring now to FIG. 5, there is shown a flow chart illustrating the process used to pre-estimate the near-future product demand. The first day to pre-estimate is selected 52, and the estimated total demand, or sales, and the day-type model for the day are retrieved 53 from the daily sales file (SLSDAY) 28. The first product planning item is retrieved 54 from the product-planning master file (DPPMST) 40. The past-product demand curve 16 for the product planning item is retrieved 55, and the number of planning cycles is determined 56 from the product-planning master file 40 and frequency tables 42 and 44. For each planning cycle 58, the rule calculations for the planning item are executed 57. The rule calculations accumulates, for the duration of the planning cycle, the past-product demand from the past product demand curve 16 for the corresponding duration of the planning cycle and represents it as a percent of past total demand. This percent will then be applied against the current daily estimated total demand to generate the near-future product demand for the planning cycle. Once this calculation has been performed, then all other rule calculations will be applied. Other rule calculations may include, for example, location-specific values (e.g., limited by the number of ovens, etc.), look-up table values, and embedded constants and calculations. For each planning cycle, steps 57 and 58 are repeated. Once all cycles for a product planning item are calculated, the pre-estimated product demand is written 61 to the pre-estimated product work file (DPPWRK) 51. For each product planning item, steps 54 through 61 are repeated. Once pre-estimating is complete for all products, control is returned to monitoring 63 the actual demand, as illustrated in the flow chart of FIG. 6.

Referring now to FIG. 6, there is shown a flow chart illustrating the process of monitoring actual demand against pre-estimated demand and revising near future-periods to reflect detected variations. The first demand period (e.g., a specific day) is selected for monitoring 65, and for each demand period, a product with a cycle that falls in that demand period is selected 67. The pre-estimated demand for the product and cycle is retrieved 69 from the pre-estimated product demand work file 51. The actual demand for this product and cycle is retrieved 71 from the sales-time file 26. The percent complete of total demand for the recent past periods is calculated 73 through the current period. The percent up or down of the product item of recent past actual demand vs. corresponding recent past pre-estimated demand is calculated 75 through the current period. The particular revision logic 79 to execute is determined 77 based upon the up or down percentage of the product determined in step 75. There are upper and lower threshold limits and a growth factor defined in the location file 49. Thus, if the percent up or down of any product exceeds the upper limit, then all remaining demand periods will be revised up by the same percentage. If the percent up or down of any product is between the upper and lower limit and the up or down percentage is positive, then no revision is performed. If the percent up or down of any product is between the upper and lower limit and the up or down percent is negative then the difference between the recent past pre-estimated demand and the corresponding actual demand through the current demand period is multiplied by the reciprocal of the percent complete of total demand determined in step 73. This value is then distributed over the remaining demand periods of the day in proportion to the pre-estimated demand for the remaining periods. If the percent up or down is below the lower limit, then all remaining demand periods will be revised down the same percentage and then revised back up by the growth factor. Steps 67 through 79 are repeated for each planning cycle that occurs within this demand period. For each remaining demand period, steps 65 through 82 are repeated. All revised product demands are displayed 83 and retained until the completion of the next demand period.

Referring now to FIG. 7, there is shown a chart of the time periods involved in operation of the present invention. Within a given day, actual data 90 regarding product sales accumulated in the recent past period 91 and in the current period 93 are used to develop a confidence factor which is then applied to the estimate of product sales in the near-future period 95 to revise the estimate 97 from the pre-estimated demand 99 for the corresponding periods, where the pre-estimated demand for products was prepared as previously described.