JP5577439B2 - Demand forecasting device, demand forecasting method and program - Google Patents

Description

  The present invention relates to a demand forecasting apparatus, a demand forecasting method, and a program for forecasting demand for goods.

  Generally, demand prediction is performed using various methods in retail stores and the like in order to prevent out of stock and excess inventory. Here, the total demand prediction of the portions is more accurate than the individual demand prediction for each portion.

Specifically, it is as described in Patent Documents 1 to 3.
In the technology described in Patent Document 1, the demand of a product is predicted for each product composed of a plurality of parts, and the demand for parts commonly used in the plurality of products is predicted based on the result. .
In the technique described in Patent Document 2, the demand for each vehicle is predicted by allocating to each vehicle based on the result of demand prediction for each vehicle classification. Each vehicle is classified into one of the vehicle classifications based on its specifications.
Moreover, in the technique described in Patent Document 3, demand prediction is performed on the consumables of the printer from a value obtained by adding up the sales results of all regions, and the result is divided at a predetermined ratio to thereby calculate the demand in each region. Is predicting.

JP 2003-242432 A JP 2001-167079 A JP 2003-233710 A

By the way, when multi-store management is performed in a retail store or the like, the prediction accuracy is often higher when the total demand at all stores is predicted than when the demand is predicted for each individual store. Especially in the case of products with a small sales volume, it is difficult to forecast demand for each store because there are days when the product can be sold and when it cannot be sold. Prediction is significantly more accurate.
However, if the total demand of all the predicted stores is allocated to each store for a product whose sales quantity is very small (for example, 5 or less per day), the number assigned to each store may be smaller than 1. For this reason, when the decimal part is rounded down, the quantity becomes 0, and when the decimal part is rounded up, there is a problem that the total demand of all stores exceeds the predicted demand value.

  The present invention has been made in view of the above points, and an object of the present invention is to provide a demand prediction apparatus, a demand prediction method, and a program capable of accurately predicting the demand amount of each store even for a product whose sales volume is small. Is to provide.

  The present invention has been made to solve the above-described problems, and one aspect of the present invention is a sales result storage unit that stores sales results of products for each store for each date, and reads from the sales result storage unit. A predicted store demand calculation unit for calculating the predicted store demand for all stores based on the sales performance of the product, and a predicted demand in each store based on the sales performance of the product read from the sales performance storage unit. The value calculated by subtracting the predicted demand for each store on the day corrected from the predicted demand for each day calculated by the predicted calculation for each store calculated by the predicted calculation for each store Based on the revised forecast value obtained by adding the carry-over value to the store-specific forecast demand amount calculated on the next day calculated by the store-specific forecast calculation unit, Forecast demand To be equal to the amount, a demand prediction unit, characterized in that it comprises a store-by-store rebate prediction calculation unit which corrects the store-by-store predicted demand for the next day, the.

  Further, according to one aspect of the present invention, in the demand prediction apparatus, the rebate prediction calculation unit for each store calculates a ratio of the corrected predicted values of the predicted demand amount for each store calculated by the prediction calculation unit for each store. Then, the predicted demand for each store is corrected in order from the store having the larger calculated ratio.

  Further, according to one aspect of the present invention, the rebate prediction calculation unit by store calculates the predicted demand amount by store for the day calculated by the prediction calculation unit by store, and the total value of the predicted demand amounts by store on the day A value obtained by subtracting the corrected predicted demand amount for each store on the current day from the predicted demand amount for each store calculated on the current day after the correction is made to be equal to the predicted total store demand on that day. Based on the calculated corrected predicted value, the corrected predicted value obtained by adding the carried forward value to the predicted demand amount by store calculated on the next day calculated by the store-specific predicted calculating unit The store-specific predicted demand amount on the next day is corrected such that the total value of the store-specific predicted demand amounts on the next day is equal to the total store predicted demand amount on the next day.

  In addition, according to one aspect of the present invention, the all-store prediction calculation unit is a predicted demand amount in all stores based on the sales results of the products read from the sales result storage unit that stores the sales results of the products for each store for each date. A step of calculating a certain all-stores predicted demand amount, and a step in which the store-specific prediction calculation unit calculates a store-specific predicted demand amount that is a predicted demand amount in each store based on the sales results of the product read from the sales result storage unit And the value calculated by subtracting the store-specific predicted demand amount for the day corrected from the store-specific forecast demand amount for the day calculated by the store-specific prediction calculation unit as a carry forward value, Based on the corrected forecast value obtained by adding the carry-over value to the forecast demand amount for each store on the day, the rebate forecast calculation unit for each store is set so that the total value of the forecast demand amounts for each store is equal to the forecast demand amount for all stores. But And correcting the store-by-store predicted demand day, a demand prediction method characterized by having a.

  In addition, according to one aspect of the present invention, all-store prediction that is a predicted demand amount in all stores based on the sales performance of the product read from the sales performance storage unit that stores the sales performance of the product for each store in the computer. A step of calculating a demand amount, a step of calculating a predicted demand amount for each store, which is a predicted demand amount in each store, based on the sales performance of the product read from the sales performance storage unit, and for each store on the calculated day Based on the corrected predicted value obtained by adding the carry forward value to the calculated predicted demand amount for the next day, the value obtained by subtracting the predicted demand amount for each store corrected from the predicted demand amount as the carry forward value, A step of correcting the predicted demand amount by store on the next day so that a total value of the predicted demand amounts by store is equal to the predicted demand amount for all stores. That.

  According to the present invention, the predicted demand for each store is corrected so that the total value of the predicted demand for each shop is equal to the predicted demand for all stores. Thereby, the total value of the predicted demand for each store does not exceed the predicted demand for all stores. Moreover, since the forecast demand amount according to store of the next day is corrected using the carry forward value, the forecast demand amount at each store can be calculated with high accuracy even for a small number of products.

It is a block diagram which shows the function structure of the demand prediction apparatus by one Embodiment of this invention. It is the schematic which shows the data structure and data example of a sales performance table which the goods sales performance memory | storage part by this embodiment memorize | stores. It is the schematic which shows the data structure and data example of a shop information table which the shop information storage part by this embodiment memorize | stores. It is the schematic which shows the data structure and data example of a merchandise information table which the merchandise information storage part by this embodiment memorize | stores. It is the schematic which shows the data structure and data example of the store prediction value table which the store prediction calculation part by this embodiment produces | generates. It is the schematic which shows the data structure and data example of the prediction value table classified by store which the prediction calculation part classified by store by this embodiment produces | generates. It is the schematic which shows the data structure and data example of a rebate prediction value table which the rebate prediction calculation part classified by store by this embodiment produces | generates. It is the schematic for demonstrating the outline | summary of the demand prediction process by this embodiment. It is a flowchart which shows the procedure of all the store predicted demand amount calculation processing by this embodiment. It is a flowchart which shows the procedure of the predicted demand amount calculation process according to store by this embodiment. It is a flowchart which shows the procedure of the rebate prediction demand amount calculation process by this embodiment. It is the schematic for demonstrating the specific example of the rebate prediction demand amount calculation process by this embodiment.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
First, the words used below are defined as follows.
The sales performance is an actual value of the sales volume of the product.
The predicted demand amount is a predicted value of the demand amount of the product obtained based on the sales performance.
The all-store predicted demand amount is a total value of the predicted demand amounts of all stores obtained from a value obtained by totaling the sales results of all target stores.
The predicted demand amount by store is the predicted demand amount of each store obtained from the sales performance of each store.
The rebate predicted demand amount is a value obtained by correcting the predicted demand amount for each store so that the total value of predicted demand amounts for each store becomes equal to the predicted demand amount for all stores. The demand forecasting device according to the present embodiment calculates the rebate forecast demand amount by correcting the forecast demand amount by store of each store so that the total value of the forecast rebate demand amount of each store becomes equal to the forecast demand amount of all stores. To do.

FIG. 1 is a block diagram showing a functional configuration of a demand prediction apparatus 1 according to an embodiment of the present invention.
The demand prediction apparatus 1 includes an input unit 11, an all-stores prediction calculation unit 12, a store-specific prediction calculation unit 13, a store-specific rebate prediction calculation unit 14, an output unit 15, and a product sales that stores the sales results of the products. It is comprised including the performance memory | storage part 16, the demand forecast value memory | storage part 17 which memorize | stores predicted demand, the store information memory | storage part 18 which memorize | stores the information of a store, and the merchandise information memory | storage part 19 which memorize | stores the information of goods. The

The input unit 11 inputs identification information (for example, a product name or a product JAN code) of a product for which demand is predicted and a prediction period (a prediction start date and a prediction end date). The all store prediction calculation unit 12 calculates the all store predicted demand amount for each date in the prediction period for the product input by the input unit 11.
The store-specific prediction calculation unit 13 calculates the store-specific predicted demand for each store on each date of the prediction period for the product input by the input unit 11. The store-specific rebate prediction calculation unit 14 calculates the store-specific prediction calculated by the store-specific prediction calculation unit 13 so that the total value of the rebate prediction demand is equal to the total store predicted demand calculated by the store prediction calculation unit 12. The demand amount is corrected to calculate the rebate predicted demand amount of each store. The output unit 15 outputs the rebate predicted demand amount calculated by the store specific rebate prediction calculation unit 14.

FIG. 2 is a schematic diagram illustrating a data structure and a data example of a sales record table stored in the product sales record storage unit 16 according to the present embodiment.
As shown in the figure, the sales result table is a two-dimensional tabular data composed of rows and columns, and includes a column for each item of product JAN (Japan Article Number) code, store ID, date, and sales result. Have. Each row of this table exists for each set of product JAN code, store ID, and date. The product JAN code is a code for identifying each product. The store ID is identification information for identifying each store individually. The date is the date when the product is sold. The sales performance is an actual value of the quantity of products sold.

FIG. 3 is a schematic diagram illustrating a data structure and a data example of a store information table stored in the store information storage unit 18 according to the present embodiment.
As shown in the figure, the store information table is two-dimensional tabular data composed of rows and columns, and includes columns for store ID, store name, location, and telephone number. Each row of this table exists for each store ID.

FIG. 4 is a schematic diagram illustrating a data structure and a data example of a product information table stored in the product information storage unit 19 according to the present embodiment.
As shown in the figure, the product information table is a two-dimensional tabular data composed of rows and columns. The product JAN code, the product name, the price, the genre, the release source, and the release date for each item. Has a column. Each row of this table exists for each product JAN code.

FIG. 5 is a schematic diagram illustrating a data structure and a data example of an all-store prediction value table generated by the all-store prediction calculation unit 12 according to the present embodiment. The all store prediction calculation unit 12 stores the all store prediction value table in the demand prediction value storage unit 17.
As shown in the figure, the all-store predicted value table is a two-dimensional tabular data composed of rows and columns, and has columns for each item of product JAN code, forecast date, and predicted total store demand. Each row of this table exists for each set of product JAN code and forecast date. The forecast date is any date in the forecast period.

FIG. 6 is a schematic diagram illustrating a data structure and a data example of a store-specific prediction value table generated by the store-specific prediction calculation unit 13 according to the present embodiment. The store prediction calculation unit 13 stores the store prediction value table in the demand prediction value storage unit 17.
As shown in the figure, the predicted value table for each store is data in a two-dimensional table format composed of rows and columns, and a column for each item of product JAN code, store ID, predicted date, and predicted demand amount by store. have. Each row of this table exists for each set of product JAN code, store ID, and prediction date.

FIG. 7 is a schematic diagram illustrating a data structure and a data example of a rebate prediction value table generated by the store rebate prediction calculation unit 14 according to the present embodiment. The store specific rebate prediction calculation unit 14 stores a rebate prediction value table in the demand prediction value storage unit 17.
As shown in the figure, the rebate prediction value table is a two-dimensional tabular data composed of rows and columns, and columns of items of product JAN code, store ID, prediction date, and rebate prediction demand amount. have. Each row of this table exists for each set of product JAN code, store ID, and prediction date.

Next, the demand prediction process by the demand prediction apparatus 1 will be described with reference to FIGS. FIG. 8 is a schematic diagram for explaining the outline of the demand prediction process according to the present embodiment.
FIG. 8A is a graph showing the transition of the predicted total store demand calculated by the predicted total store calculation unit 12. The horizontal axis shown in this figure is the forecast date, and the vertical axis is the predicted demand for all stores. FIG. 8B is a graph showing the transition of the predicted demand amount by store in a specific store (for example, store A). The horizontal axis shown in this figure is the predicted date, and the vertical axis is the predicted demand amount by store. FIG. 8C is a graph showing the transition of the rebate predicted demand amount calculated by the rebate prediction calculation unit 14 for each store. The horizontal axis shown in this figure is the forecast date, and the vertical axis is the forecast demand. Moreover, the solid line shown in this figure is the total store predicted demand amount, and the broken line is the rebate predicted demand amount in the store A. The rebate prediction calculation unit 14 for each store corrects the predicted demand amount for each store so that the total sum of the predicted rebate demand amounts for each store becomes equal to the predicted demand amount for all stores. Is calculated. Details of the processing of the rebate prediction calculation unit 14 by store will be described later.

FIG. 9 is a flowchart showing a procedure of the all-store predicted demand calculation process according to this embodiment.
First, in step S <b> 101, the whole store prediction calculation unit 12 reads a sales record corresponding to the product input from the input unit 11 from the sales result table.
Next, in step S <b> 102, the whole store prediction calculation unit 12 sets the prediction date as the prediction start date input from the input unit 11.
Next, in step S103, the all store prediction calculation unit 12 calculates the all store predicted demand amount on the prediction date based on the sales record read in step S101. Specifically, first, the store prediction calculation unit 12 obtains parameters a, b, and c by fitting a model formula defined by a logistic curve by a sales result and a least square method. The model formula is defined by the following formula (1), for example. However, f (t) is the total store demand amount, and t is the forecast date. And the all store prediction calculation part 12 calculates the all store predicted demand amount in a prediction date using the calculated | required parameter and Formula (1).

  f (t) = a / (1 + b · exp (−ct)) (1)

  In the present embodiment, a model formula based on a logistic curve is used. However, for example, a model formula based on a Poisson distribution may be used. Alternatively, a moving average of sales results, a regression equation obtained as a result of regression analysis using the sales results and product information as variables, and other statistically obtained prediction formulas may be used. Moreover, depending on the case, you may correct | amend all the store predicted demand amounts using the incidental information, such as the weather of the day, temperature, etc., the information which shows that it is a sale period.

Next, in step S <b> 104, the store prediction calculation unit 12 determines whether or not the prediction date is the prediction end date input from the input unit 11. If the predicted date is the predicted end date, the process proceeds to step S106. On the other hand, if the predicted date is not the predicted end date, the process proceeds to step S105.
In step S105, the all-stores prediction calculation unit 12 sets the prediction date as the day after the prediction date, and returns to step S103.
On the other hand, in step S106, the store prediction calculation unit 12 records the calculated store predicted demand amount in the store prediction value table and stores it in the demand prediction value storage unit 17.

FIG. 10 is a flowchart illustrating the procedure of the predicted demand calculation process for each store according to the present embodiment.
First, in step S201, the store-specific prediction calculation unit 13 selects a store for which the store-specific predicted demand is calculated.
Next, in step S202, the store prediction calculation unit 13 reads the sales record corresponding to the store selected in step S201 and the product input by the input unit 11 from the sales record table.
Next, in step S <b> 203, the store-specific prediction calculation unit 13 sets the prediction date as the prediction start date input by the input unit 11.

  Next, in step S204, the store-specific prediction calculation unit 13 calculates the store-specific predicted demand amount on the prediction date based on the sales record read in step S202. Specifically, the store-specific prediction calculation unit 13 sets the moving average value of the sales performance as the store-specific predicted demand amount. Alternatively, the predicted demand amount for each store may be calculated using a predetermined probability distribution. In the present embodiment, the store-specific predicted demand amount is calculated using a method different from the all-store prediction calculation unit 12, but the store-specific predicted demand amount may be calculated using the same method.

Next, in step S <b> 205, the store-specific prediction calculation unit 13 determines whether or not the prediction date is a prediction end date input by the input unit 11. If the predicted date is the predicted end date, the process proceeds to step S207. On the other hand, if the predicted date is not the predicted end date, the process proceeds to step S206.
In step S206, the store-specific prediction calculation unit 13 sets the prediction date as the day after the prediction date and returns to step S204.
On the other hand, in step S207, the store-specific prediction calculation unit 13 determines whether the store-specific predicted demand amount has been calculated for all stores. When the predicted demand for each store is calculated for all stores, the process proceeds to step S209. On the other hand, if the predicted demand for each store has not been calculated for all stores, the process proceeds to step S208.
In step S208, the store-specific prediction calculation unit 13 selects the next store and returns to step S202.
On the other hand, in step S <b> 209, the store-specific prediction calculation unit 13 records the calculated store-specific predicted demand amount in the store-specific prediction value table and stores it in the demand prediction value storage unit 17.

FIG. 11 is a flowchart illustrating a procedure of rebate predicted demand calculation processing according to the present embodiment.
First, in step S <b> 301, the store-specific rebate prediction calculation unit 14 sets the prediction date as the prediction start date input from the input unit 11.
Next, in step S <b> 302, the store-specific rebate prediction calculation unit 14 reads out the all-store predicted demand amount and the store-specific predicted demand amount on the prediction date from the demand prediction value storage unit 17.

  Next, in step S303, the rebate prediction calculation unit 14 for each store corrects the read predicted demand amount for each store so that the total value of the predicted rebate demand amount becomes equal to the read all store predicted demand amount. Calculate the predicted demand. Specifically, first, the rebate prediction calculation unit 14 for each store calculates the ratio of the predicted demand for each store. And the rebate prediction calculation part 14 classified by store distributes all the store predicted demand quantities in an order from a shop with a larger ratio based on the ratio of the calculated forecast demand quantity classified by store.

Next, in step S304, the rebate prediction calculation unit 14 for each store determines whether or not the prediction date is a prediction end date. If the predicted date is the predicted end date, the process proceeds to step S309. On the other hand, if the predicted date is not the predicted end date, the process proceeds to step S305.
In step S305, the rebate prediction calculation unit 14 for each store sets the prediction date as the day after the prediction date.
Next, in step S <b> 306, the store-by-store rebate prediction calculation unit 14 reads out the all-stores predicted demand amount and the store-by-store predicted demand amount on the prediction date from the demand prediction value storage unit 17.

  Next, in step S307, the rebate prediction calculation unit 14 for each store calculates a corrected prediction value by the following equation (2).

  Corrected forecast value (prediction date) = predicted demand amount by store (prediction date) + (predicted demand amount by store (prediction date-1) −rebate forecast demand amount (prediction date-1)) (2)

Then, in step S308, the rebate prediction calculation unit 14 for each store starts from the store having the calculated ratio of the corrected predicted values in order from the store having the larger ratio of the predicted rebate demand amount to the read all store predicted demand amount. The corrected predicted value is corrected to calculate a rebate predicted demand amount, and the process returns to step S304.
On the other hand, in step S <b> 309, the rebate prediction calculation unit 14 for each store records the calculated rebate predicted demand amount in the rebate predicted value table and stores it in the demand predicted value storage unit 17.
Next, in step S310, the rebate predicted demand amount recorded by the output unit 15 is displayed on a display unit (not shown) or the like.

  FIG. 12 is a schematic diagram for explaining a specific example of the rebate predicted demand calculation processing according to the present embodiment. In the example shown in this figure, the rebate predicted demand amount of the store A, the store B, and the store C in the prediction period from April 1 to April 3 is calculated. Moreover, the carry-over value in this figure is a value obtained by subtracting the rebate predicted demand amount from the store-specific predicted demand amount.

  First, the rebate forecast calculation unit 14 for each store calculates a rebate forecast demand amount on April 1st. The predicted demand for all stores on April 1 is 3, the predicted demand for each store of store A is 1, the predicted demand for each store of store B is 1, and the predicted demand for each store of store C is 1. It is. First, the rebate prediction calculation unit 14 for each store calculates the ratio of the predicted demand amount for each store. The calculated ratio is store A: store B: store C = 1: 1: 1. Moreover, since the value which totaled the predicted demand amount according to store of stores A to B is 3, which is equal to the predicted demand amount for all stores, the rebate forecast calculation unit 14 according to store sets the rebate predicted demand amount of each store as 1. To do.

Next, the rebate forecast calculation unit 14 for each store calculates a rebate forecast demand amount on April 2. The predicted demand for all stores on April 2 is 5, the predicted demand for each store of store A is 1.5, the predicted demand for each store of store B is 1.3, and the predicted demand for each store of store C is 1.3. The demand is 1.2.
First, the rebate prediction calculation unit 14 for each store calculates a corrected predicted value by adding the carry-over value of each store and the predicted demand amount for each store. Since the carry-over value of stores A to C is 0, the corrected predicted value of stores A to C is the same as the predicted demand amount for each store. Next, the rebate prediction calculation unit 14 for each store calculates the ratio of the corrected predicted values for each store. The calculated ratio is store A: store B: store C = 15: 13: 12, and the total calculated ratio is 40. Here, since the ratio of the store A is the largest, the rebate prediction calculation unit 14 for each store first sets the rebate predicted demand amount of the store A to 2 (5 (all store predicted demand amount) × 0.375 (store A Ratio 15 / total 40) = 1.875 rounded off the decimal point). Next, since the ratio of the store B is large, the store-specific rebate prediction calculation unit 14 sets the rebate predicted demand amount of the store B to 2 (5 (all store predicted demand amount) × 0.325 (the ratio 13 / ratio of the store B). 40) = 1.625 (rounded off to the nearest decimal point). Finally, the rebate prediction calculation unit 14 for each store sets the rebate predicted demand amount of the store C to 5 (all store predicted demand amount) -4 ( (Total value of rebate forecast demand amount of store A and store B) = 1.

Next, the rebate forecast calculation unit 14 for each store calculates a rebate forecast demand amount on April 3rd. The predicted demand for all stores on April 3 is 7, the predicted demand for each store in store A is 2.5, the predicted demand for each store in store B is 2, and the predicted demand for each store in store C is 2. Is 2. Further, the carry-over value of the store A is −0.5, the carry-over value of the store B is −0.7, and the carry-over value of the store C is 0.2. For this reason, the corrected predicted value of the store A is 2.5 + (− 0.5) = 2, the corrected predicted value of the store B is 2 + (− 0.7) = 1.3, and the corrected predicted value of the store C The value is 2 + 0.2 = 2.2. Moreover, the ratio of the corrected predicted value of each store is store A: store B: store C = 20: 13: 22, and the total of the calculated ratios is 55. Here, since the ratio of the store C is the largest, the rebate prediction calculation unit 14 for each store sets the rebate predicted demand amount of the store C to 3 (7 (all store predicted demand amount) × 0.4 (the ratio 22 of the store A). / Total ratio 55) = value obtained by rounding off the fractional part of 2.8.
Next, since the ratio of the store A is large, the rebate prediction calculation unit 14 for each store calculates the rebate predicted demand amount of the store A by 2 (7 (all store predicted demand amount) × 0.36 (the ratio 20 / ratio of the store A). (55) = value obtained by rounding down the decimal part of 2.5). Finally, the rebate prediction calculation unit 14 for each store sets the rebate predicted demand amount of the store B to 7 (total predicted demand amount of all stores) -5 (total value of predicted rebate demand amounts of the store C and the store A) = 2. To do.

  In addition, when rebating from the ratio of each store, the fractional part that does not meet the unit quantity may be rounded off or rounded off. In that case, the predicted demand amount by store (or revised forecast value) And the difference between rebate forecast demands as carry forward values.

  Thus, according to the present embodiment, the rebate predicted demand amount is calculated by correcting the predicted demand amount for each store so that the total value of the rebate predicted demand amount becomes equal to the total store predicted demand amount. As a result, the total value of the predicted rebate demand does not exceed the predicted demand for all stores. In addition, considering the number of decimal places, the rebate forecast demand amount is calculated by adding the carry-forward value (predicted demand amount by store-rebate forecast demand amount) to the forecast demand amount by store on the next day. Even if it exists, the predicted demand amount in each store can be calculated accurately. In addition, it is possible to appropriately distribute products with little demand at all stores, such as scarce products, to each store without any bias. In particular, even if there is a sudden change in the sales performance of an individual store, it is possible to make a prediction that excludes the influence.

  In the present embodiment, the store-by-store predicted demand amount for each date is rebated for each store. However, for example, the store-by-store predicted demand amount may be rebated for a predetermined period. In that case, the ratio of each store such that the sum of each store becomes 1, such as the sales volume of the total store for a predetermined period and the total sales amount, can be used.

Further, the demand prediction process may be performed by recording a program for realizing each step shown in FIGS. 9 to 11 on a computer-readable recording medium. Here, the “computer system” may include an OS and hardware such as peripheral devices.
Further, the “computer system” includes a homepage providing environment (or display environment) if a WWW system is used.
The “computer-readable recording medium” means a flexible disk, a magneto-optical disk, a ROM, a writable nonvolatile memory such as a flash memory, a portable medium such as a CD-ROM, a hard disk built in a computer system, etc. This is a storage device.

Further, the “computer-readable recording medium” means a volatile memory (for example, DRAM (Dynamic DRAM) in a computer system that becomes a server or a client when a program is transmitted through a network such as the Internet or a communication line such as a telephone line. Random Access Memory)), etc., which hold programs for a certain period of time.
The program may be transmitted from a computer system storing the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the “transmission medium” for transmitting the program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line.
The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, and what is called a difference file (difference program) may be sufficient.

As described above, the embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to the above, and various design changes and the like can be made without departing from the scope of the present invention. It is possible to
For example, in the present embodiment, the rebate predicted demand amount is calculated by correcting the predicted demand amount for each store, but even if the rebate predicted demand amount is calculated from the average value of the sales performance in a certain period in the past. Good.

DESCRIPTION OF SYMBOLS 1 ... Demand prediction apparatus 11 ... Input part 12 ... All store prediction calculation part 13 ... Prediction calculation part classified by store 14 ... Rebate prediction calculation part classified by store 15 ... Output part 16 ... Merchandise sales performance memory | storage part 17 ... Demand prediction value memory | storage part 18 ... Store information storage unit 19 ... Product information storage unit

Claims (5)

A sales performance storage unit that stores the sales performance of products for each store for each date;
An all-store prediction calculation unit that calculates an all-stores predicted demand amount that is a predicted demand amount in all stores based on the sales results of the product read from the sales result storage unit;
A store-specific prediction calculation unit that calculates a store-specific predicted demand amount that is a predicted demand amount in each store based on the sales results of the product read from the sales result storage unit;
The next day calculated by the store-specific prediction calculation unit using a value obtained by subtracting the store-specific predicted demand amount corrected from the store-specific predicted demand amount of the day calculated by the store-specific prediction calculation unit as a carry forward value Based on the revised forecast value obtained by adding the carry-over value to the store forecast demand amount, the store forecast demand on the next day so that the total value of the store forecast demand amounts becomes equal to the store forecast demand amount. Rebate prediction calculation unit by store for correcting the amount;
A demand prediction apparatus comprising:   The rebate calculation unit for each store calculates a ratio of the corrected predicted value of the predicted demand amount for each store calculated by the prediction calculation unit for each store, and the predicted demand amount for each store in descending order of the calculated ratio. The demand prediction apparatus according to claim 1, wherein: The rebate prediction calculation unit by store is
After correcting the forecasted demand for each day calculated by the forecasted calculation unit for each store so that the total value of the forecasted demand for each day of the store is equal to the forecasted demand for all stores on that day, The next day calculated by the store-specific forecast calculation unit using the value calculated by subtracting the corrected store-specific forecast demand amount on the current day as the carry-over value from the store-specific forecast demand amount calculated by the forecast calculation unit The corrected predicted value obtained by adding the carry-over value to the predicted demand amount of each store is calculated, and the total value of the predicted demand amounts by store on the next day is calculated based on the calculated corrected predicted value on the next day. The demand prediction apparatus according to claim 1 or 2, wherein the store-specific predicted demand amount on the next day is corrected so as to be equal to the total store predicted demand amount. A step of calculating an all-stores predicted demand amount, which is a predicted demand amount in all stores, based on the sales results of the products read from the sales result storage unit that stores the sales results of the products for each store by date. When,
A step of calculating a predicted demand amount by store, which is a predicted demand amount in each store, based on the sales result of the product read from the sales result storage unit,
The next day calculated by the store-specific prediction calculation unit using a value obtained by subtracting the store-specific predicted demand amount corrected from the store-specific predicted demand amount of the day calculated by the store-specific prediction calculation unit as a carry forward value Based on the corrected predicted value obtained by adding the carry-over value to the predicted demand amount for each store, the rebate prediction calculation unit for each store performs the following so that the total value of the predicted demand amounts for each store becomes equal to the predicted total demand amount for all stores. Correcting the predicted demand for each store on the day of
A demand forecasting method characterized by comprising: On the computer,
A step of calculating an all-stores predicted demand amount that is a predicted demand amount in all stores based on the sales results of the products read from the sales result storage unit that stores the sales results of the products for each store for each date;
Calculating a predicted demand amount for each store, which is a predicted demand amount in each store, based on the sales results of the product read from the sales result storage unit;
The carry-over value is added to the calculated predicted demand amount on the next day, with the value obtained by subtracting the predicted demand amount by store corrected on the current day from the calculated predicted demand amount on the current day as the carry-over value. Correcting the predicted demand for each store on the next day based on the corrected predicted value so that the total value of the predicted demand for each store is equal to the predicted demand for all stores;
A program for running

Images