JP2016012321A - Demand prediction device and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a demand prediction device capable of more quickly acquiring demand prediction results of a plurality of shops, and further to provide a program.SOLUTION: A demand prediction device comprises: a first storage region storing purchase result data of a plurality of shops; generation means generating purchase result data of one shop by summing up the purchase result data of the plurality of shops stored in the first storage region; a second storage region storing the purchase result data of one shop generated by the generation means; demand prediction means making a demand prediction calculation on the basis of the purchase result data of one shop stored in the second storage region; and results output means outputting demand prediction calculation results of the demand prediction means.

Description

  Embodiments described herein relate generally to a demand prediction apparatus and a program.

  Conventionally, POS (Point of Sales) systems have become widespread, and it has become possible to easily collect purchase performance (sales) data for each product. And the demand forecast according to goods, such as a retail store, based on the purchase performance data for every goods has begun to be used. For example, a demand forecast obtained by analyzing purchase record data can be used to suppress a loss of unsold products due to an excessive order quantity or an opportunity loss due to an out of order product due to an insufficient order quantity. Then, a technology is being proposed that combines demand forecasting with the digitization of business-to-business transactions using EDI (Electronic Data Interchange) to develop an automatic ordering system.

  As a demand forecasting technique that can be used for an automatic ordering system, there is a known multiple regression analysis. And a demand prediction can be performed for every goods by using a multiple regression analysis.

  In demand forecast using conventional multiple regression analysis, it has been tried to use sales price, sales form implementation period, display position, number of displays, advertising method, etc. as variable factors as the variable factors of the number of purchases that are objective variables. Yes. In addition, in consideration of the mutual influence between a product and its related products, attempts have been made to use the number of purchases of each product as a variable factor.

By the way, in the demand forecast using the conventional multiple regression analysis, in order for the store ordering person to know the demand forecast of his / her store, the purchase forecast data of his / her store is applied to the formula of the demand forecast, and the predicted number of purchases To determine the order quantity of the product.
On the other hand, in the headquarters that manages a plurality of stores, buyers may order products that are common to the plurality of stores and keep the purchase price low.
In this case, the buyer has to make a demand forecast for each store and grasp the predicted number of purchases of all stores.
However, if demand prediction is performed for each store, there is a problem because it takes time to acquire all the demand prediction results as the number of stores increases. As the number of stores increases, the need for batch ordering increases, so improvement is desired.

  The problem to be solved by the present invention is to provide a demand prediction device and a program capable of acquiring the result of demand prediction of a plurality of stores more quickly.

  The demand prediction apparatus according to the embodiment includes a first storage area that stores purchase record data of a plurality of stores and a purchase record data of the plurality of stores stored in the first storage area. A generating unit that generates purchase record data of a store, a second storage area that stores purchase record data of the one store generated by the generating unit, and the one store that is stored in the second storage area Demand forecasting means for performing demand forecast calculation based on the purchase record data of the customer, and result output means for outputting the result of the demand forecast calculation of the demand forecasting means.

FIG. 1 is a schematic explanatory diagram of a demand prediction system according to the present embodiment. FIG. 2 is a block diagram illustrating a configuration of the demand prediction system according to the embodiment. FIG. 3 is a module configuration diagram of the computer. FIG. 4 is a block diagram showing the configuration of the headquarter system including the demand prediction device. FIG. 5 is a data configuration diagram of store data. FIG. 6 is a configuration diagram of aggregate store data obtained by aggregating store data. FIG. 7 is store data shown as an example of schedule information. FIG. 8 shows an example of a configuration diagram of the plan editing screen of the headquarters. FIG. 9 is a flowchart of processing relating to data collection and demand prediction simulation performed at night in the demand prediction apparatus. FIG. 10 is a flowchart of a demand prediction simulation process performed between the headquarters special sales planning terminal and the demand prediction device. FIG. 11 is a display example of the plan editing screen after re-simulation. FIG. 12 is a diagram showing a state when the confirmation screen is displayed in a pop-up. FIG. 13 is a display screen example of demand forecast data for each store. FIG. 14 is a display example of the store-specific prediction screen including the distribution amount.

FIG. 1 is a schematic explanatory diagram of a demand prediction system according to the present embodiment.
Store data D1 to store data DK (K = 1, 2, 3,...) Shown in FIG. 1 are collected over a certain period for each store (store 1, store 2,..., Store K). Purchase data is shown.

  Purchasing performance data for each store includes condition data that does not depend on the sales region (non-regional data), such as the selling price / display of the product, and condition data (regional data) that varies depending on the sales region, such as promotional leaflets, weather, and temperature. And the number of purchases indicating the actual number of items sold under those conditions.

The total store data DT is obtained by integrating and totaling store data D1 to store data DK of all stores (store 1 to store K) as indicated by arrows f1 to fK. This processing is performed at night.
The aggregate store data DT is a virtual store in which a summary store that is regarded as one store is stored in the memory, and the store holds the store data.

  The buyer S of the headquarters launches a demand forecast system screen (demand forecast setting screen, plan edit screen, etc.) G, changes the selling price of one product, and executes demand forecast calculation (demand forecast simulation). Then, the demand prediction device first performs a demand prediction calculation based on the purchase record data (aggregate store data DT) of the aggregation store, and outputs the calculation result to the screen G of the demand prediction system (ST1).

  While outputting the first result to the screen G of the demand prediction system, the demand prediction device performs a demand prediction calculation process based on purchase history data unique to each store for each store. And when this process is implemented with respect to all the store data (store data D1-store data DK) and the process is completed, the notification which shows the completion of the process is output to the screen G of a demand prediction system. After the output of the notification, when the buyer S at the headquarters performs an operation of switching the display from the demand prediction result of the aggregate store to the demand prediction result for each store, the demand prediction device is connected to each store (store 1, store 2,... , Store K) demand prediction result is output to screen G of the demand forecasting system (ST2).

Below, the specific structure of the demand prediction system shown in FIG. 1 is demonstrated.
FIG. 2 is a block diagram illustrating a configuration of the demand prediction system 1 according to the embodiment. As shown in FIG. 2, the demand forecasting system 1 has a plurality of headquarters special sale plans for a headquarters system formed by a plurality of computers 3 electrically connected to each other via a wired or wireless communication line 2. A terminal 124 and a store ordering terminal 125 provided in each store are connected via a wired or wireless communication line (for example, the Internet or a LAN) 5.

  In the demand prediction system 1 of the present embodiment, one or more computers 3 function as an application server or a database server that provides a predetermined service (application). Further, the headquarters special sales planning terminal 124 and the store ordering terminal 125 function as clients that enjoy the service. In the demand prediction system 1, a service (application) can be provided in the form of SaaS (Software as a Service), for example. The demand prediction system 1 may be in the form of a server client.

  In addition, a store server 120 that connects a plurality of POS (Point of Sales) terminals 126 is connected to the head office system via a wired or wireless communication line (for example, the Internet or a LAN) 5.

  The computer 3, the headquarters special sales planning terminal 124, and the store ordering terminal 125 as described above are general personal computers. The headquarters special sales planning terminal 124 and the store ordering terminal 125 may be tablet terminals. Here, the computer 3 will be described as an example.

  FIG. 3 is a module configuration diagram of the computer 3. As shown in FIG. 3, the computer 3 includes a CPU (Central Processing Unit) 101 that performs information processing, a ROM (Read Only Memory) 102 that is a read-only memory that stores BIOS, and a RAM that stores various data in a rewritable manner. (Random Access Memory) 103 is provided.

  In addition, the computer 3 functions as various databases and stores information using an HDD (Hard Disk Drive) 104 and a storage medium 110 which are storage units for storing various programs, and distributes information to the outside. Media reading device 105 such as a DVD drive for obtaining information, communication control device 106 for transmitting information to other external devices via each communication line, LCD for displaying processing progress and results to the operator A display unit 107 such as (Liquid Crystal Display) and an input unit 108 that is a pointing device such as a keyboard and a mouse for an operator to input commands and information to the CPU 101 are provided.

  In the computer 3, data transmitted / received between the above-described units is arbitrated by the bus controller 109.

  In such a computer 3, when an operator turns on the power, the CPU 101 activates a program called a loader in the ROM 102, reads a program for managing the hardware and software of the computer called OS (Operating System) from the HDD 104 into the RAM 103, and Start the OS. Such an OS activates a program, reads information, and stores information in response to an operator's operation. As a representative OS, Windows (registered trademark) and the like are known. These operation programs running on the OS are called application programs. The application program is not limited to one that runs on a predetermined OS, and may be one that causes the OS to execute some of the various processes described below, or constitutes predetermined application software, an OS, or the like. It may be included as part of a group of program files.

  That is, the personal computer functions as the headquarters special sales planning terminal 124, the store ordering terminal 125, and the computer 3 (application server or database server) that function as clients, depending on the difference of application programs stored in the HDD 104.

  For example, the head office special sales planning terminal 124 and the store ordering terminal 125 are installed with a web browser and software provided to the web browser using a plug-in format as an application program. A typical plug-in is Microsoft Silverlight (registered trademark). Microsoft Silverlight is a framework that, after installation, displays animations and graphics on the browser and plays audio and video.

  In general, an application program installed in the HDD 104 of the headquarters special sales planning terminal 124 is recorded in a portable storage medium 110 such as an optical information recording medium such as a CD-ROM or DVD or a magnetic medium such as an FD. Then, the operation program recorded in the storage medium 110 is installed in the HDD 104. Furthermore, the application program may be imported from the outside via the communication control device 106 and installed in the HDD 104, for example.

In the computer 3, when an application program operating on the OS is started, the CPU 101 executes various arithmetic processes according to the application program, and controls each unit intensively.
In the following, a demand prediction process, which is a characteristic process of the present embodiment, will be described among various arithmetic processes executed by each CPU 101 of one or a plurality of computers 3 constituting the headquarters system using an application program.

<< Configuration of demand forecasting device and its peripheral system >>
FIG. 4 is a block diagram illustrating the configuration of the headquarter system including the demand prediction apparatus according to the present embodiment. As shown in FIG. 4, each CPU 101 of one or a plurality of computers 3 constituting the headquarters system executes an application program, whereby a demand prediction device 31, a data collection / conversion system 32, a scheduled data collection / conversion system. 33 (special sale planning support system 330, order quantity determination support system 331, etc.) and order system 34 are realized.
In the present embodiment, the special sales planning support system 330 and the order quantity determination support system 331 are shown outside the demand prediction device 31. However, the special sales planning support system 330 and the order quantity determination support system 331 are subsystems for providing the service of the demand prediction device 31 to the user terminal. Therefore, both or any one of the special sales planning support system 330 and the order quantity determination support system 331 may be appropriately included in a part of the demand prediction device 31.

<Demand prediction device>
The demand prediction device 31 predicts the demand (for example, the number of purchases) of a product with respect to the sales state scheduled by the seller, such as the selling price / display of the product, at a headquarters or a store such as a retail business by simulation. The calculation formula used for the simulation is, for example, a constant term indicating the number of purchases that can be regarded as not changing in time for about several weeks, months, or one year, and a portion that changes more quickly with respect to the constant term. It can be indicated by a term including a variation factor indicating the number of purchases and a coefficient.

By using this calculation formula, a purchase model for the customer is built based on past information (past purchase record data) that shows the sales status and purchase results at that point in time from the present time to a predetermined point in the past. Is done.
Next, the calculation formula is shown.
In the present embodiment, a case will be described in which there are N types of products that constitute a product group (a product and its related products). N types of products are denoted as product 1, product 2,..., Product n,.

  Further, the variation factors for the product n are indicated as Xn1, Xn2,..., Xnm,..., XnM (m = 1, 2,. The variable factor Xnm may change the number of purchases, such as standard price, special price (selling price), discount amount, display state, leaflet placement, day of the week, holidays, maximum temperature, minimum temperature, precipitation, etc.

Further, the number of purchases for the product n (n = 1, 2,..., N) is Yn. That is, the number of purchases of product 1, product 2,..., Product n,..., Product N is indicated as purchase number Y1, Y2,.
In this embodiment, considering that the number of purchases of each product interacts with the number of purchases of other products, it is assumed that Y1, Y2,..., Yn,. The following formula (1) shows the calculation formula.

上記の式（１）において、ａn1Ｙ1＋ａn2Ｙ2＋…＋ａn(n-1)Ｙn-1＋ａn(n+1)Ｙn+1＋…＋ａnNＹNは関連商品の購買数による影響を反映した項である。
式（１）において、ａn1,…,ａn(n-1),ａn(n+1),…,ａnN,ｂn0（定数項）,…,ｂnM、を回帰係数とする。本実施形態では、回帰係数を定数項ｂn0を含むものとして説明する。

In the above formula (1), an1Y1 + an2Y2 + ... + an (n-1) Yn-1 + an (n + 1) Yn + 1 + ... + anNYNN is a term reflecting the influence of the number of related products purchased.
In equation (1), an1,..., An (n-1), an (n + 1),..., AnN, bn0 (constant terms),. In the present embodiment, the regression coefficient is described as including a constant term bn0.

<Data collection / conversion system>
The data collection / conversion system 32 includes data indicating the number of purchases of a plurality of products in a store from the present time to a predetermined time in the past, selling price data, sales promotion data, ordering data, inventory data, weather data, day data, and region The past information such as event data is collected from the store server 120 of each store. The data collection / conversion system 32 converts a plurality of various data (collected past information) into a predetermined data structure and transmits the data to the past information database 310 of the demand prediction device 31. The data collection / conversion system 32 includes various data collection systems such as a known POS system, an ordering system, a delivery product inspection system, a weather data reception system, a local event input system, and a data conversion system. ing.

<Planned data collection / conversion system>
The scheduled data collection / conversion system 33 includes planned sales price data, planned sales data, planned arrival data, planned weather data, day of week data, planned area event data, etc. for a plurality of products from the present time to a predetermined future time. Collect future schedule information. The scheduled data collection / conversion system 33 converts the collected various data into a predetermined data structure, and transmits the data to the future information database 311 of the demand prediction device 31.
The scheduled data collection / conversion system 33 includes a special sale planning support system 330, an order quantity determination support system 331, and the like.

  The special sale planning support system 330 is a system constructed to support special sale planning. It is interposed between the demand prediction device 31 and the headquarters special sales planning terminal 124, and provides an interface for processing the special sales planning to the headquarters special sales planning terminal 124.

  The headquarters special sales planning terminal 124 is a terminal device used by a person in charge of special sales planning at the headquarters or a buyer at the headquarters as a user. The headquarters special sales planning terminal 124 accepts input of future schedule information such as a product code (single product code), a special sale price (sale price), a special sale date, a flyer publication, etc. of a special sale planning product group, and the input data is used as a special sale planning support system. 330.

  The headquarters special sales planning terminal 124 accepts various input operations (data editing operation, re-simulation execution operation, etc.) for performing a demand prediction simulation, and transmits the input data, operation signals, and the like to the special sales planning support system 330. . The headquarters special sales planning terminal 124 receives data indicating demand prediction simulation results (demand prediction data) from the special sales planning support system 330 as a result of various input operations.

  The order quantity determination support system 331 is a system constructed to support the determination of the order quantity. It intervenes between the demand prediction device 31 and the store ordering terminal 125 used by the person in charge of ordering the store, and provides an interface for determining the number of orders to the store ordering terminal 125.

  The order quantity determination support system 331 outputs a demand prediction simulation execution command from the store ordering terminal 125 to the input reception unit 317. In addition, the order quantity determination support system 331 transmits the result of the demand prediction simulation from the input reception unit 317 to the store order terminal 125. In the present embodiment, it is assumed that the order quantity determination support system 331 controls the input reception unit 317 to perform data and commands transmitted from the external device to the input reception unit 317. For this reason, it is set as the aspect which makes the order number determination support system 331 output the data and command which were received in the special sales plan support system 330.

The order quantity determination support system 331 receives the order instruction output from the special sales planning support system 330 and outputs it to the order system 34.
The store ordering terminal 125 is a terminal device for a store ordering person to determine the number of items ordered in the store. The store ordering terminal 125 performs a demand prediction simulation and determines the number of orders based on the result. Input data and operation signals for simulation received at the store order terminal 125 are transmitted to the order quantity determination support system 331. The demand forecast data as a result of the simulation is received from the order quantity determination support system 331.

  The ordering system 34 is a system that requests an order for a product from a manufacturer that manufactures or sells the product according to the order number determined by the order number determination support system 331.

<< Configuration and processing of demand forecasting device >>
<Configuration of demand forecasting device>
The demand prediction device 31 includes a past information database 310 having past information and a future information database (DB) 311 having future information in a storage unit. The future information database 311 includes schedule information 311A and prediction information 311B as future information.
The past information database 310 has past purchase record data (store data D1a to store data DKa) of each store as past information in a predetermined storage area of the storage unit. Further, the schedule information 311A of the future information database 311 has store data D1b to store data DKb as schedule information of each store in a predetermined storage area of the storage unit.

The past information database 310 further has aggregate store data DTa as past information in a predetermined storage area of the storage unit.
The future information database 311 further has aggregate store data DTb as schedule information in a predetermined storage area of the storage unit.

  The total store data DTa is total data of all stores obtained by totaling store data D1a to store data DKa including past information of each store. This total shop data DTa is generated by a totaling unit (generation unit) 312.

  The totaling unit 312 extracts the store data D1a, D2a,..., DKa of each store, integrates and totals them, and stores them in the past information database 310 as store data (total store data DTa) of one store called a total store. To do.

Here, the generation process of the aggregate store data DTa will be described.
FIG. 5 is a data configuration diagram of store data D1a shown as an example of store data D1a to DKa.

  Store data D1a in FIG. 5 includes “store code” 601, “date” 611, “day of the week” 612, “holiday” 613, “maximum temperature” 621, “minimum temperature” 622, “precipitation” 623, “ "Line code" 631, "Class code" 632, "Product code" 633, "Standard price" 641, "Sales price" 642, "Discount" 643, "Flyer publication" 644, "Display status" 651, "Daily purchase" It has a data item such as “number” 662.

“Store code” 601 is an item indicating a store identification code. Since the data corresponds to the store 1, “001” is set in the “store code” 601.
“Year / Month / Day” 611, “Day of the Week” 612, and “Holiday” 613 are items indicating “Year / Month / Day”, “Day of the Week”, and “Holiday” of the purchase date, respectively. For “day of the week” 612 and “holiday” 613, the flag “1” is set when applicable.

“Maximum temperature” 621, “Minimum temperature” 622, and “Precipitation” 623 are items indicating weather data on the purchase date.
“Line code” 631 and “class code” 632 are items indicating the classification of purchased products. The product category is classified into “department”, “line code”, and “class code” from the top to the bottom. Since “department” is the highest level and is not classified elsewhere, only “line code” and “class code” are set here.

“Product code” 633 is an identification code of the purchased product.
The “standard price” 641 is an item indicating the product price at the time of standard sales.
The “sale price” 642 is an item indicating a product price at the time of special sale.

The “discount amount” 643 is an item indicating a discount amount from the standard price. Here, an amount obtained by subtracting the selling price from the standard price is set.
“Flyer publication” 644 is an item indicating a sales promotion method. Here, a flag “1” is set when a leaflet is placed.

The “display state” 651 is an item indicating the sales state of the product. Here, a number indicating a preset display pattern is set.
The “daily purchase quantity” 662 is an item indicating the actual number of purchases on the date set in the “year / month / day” 611. Here, the number of daily purchases of one product is set.

Below each data item, a part of the data accumulated in the store 1 is shown.
One record indicates the purchase record of one product for each purchase date. In the example of the figure, as indicated by the data of “Date” 611 and the data of “Product Code” 633, first, records corresponding to the number of products indicating the purchase results of January 16, 2014 are shown, and subsequently The records corresponding to the number of products indicating the purchase record of January 17, 2014 on the next day are shown. After that, although omitted, data up to the date of collecting this data or data up to the vicinity thereof are included.

  In the store data D1a, the non-regional data includes “date of year” 611, “day of the week” 612, “holiday” 613, “standard price” 641, “sale price” 642, “discount amount” 643, “display state” ”651. Further, the regional data is data of “maximum temperature” 621, “minimum temperature” 622, “precipitation” 623, and “flyer publication” 644.

FIG. 6 is a configuration diagram of aggregate store data DTa obtained by aggregating store data D1a to store data DKa.
The total store data DTa in FIG. 6 is obtained by extracting and totaling data common to the store data D1a to the store data DKa to obtain purchase record data of the total store.

Specifically, the aggregate store data DTa is generated as follows.
For the store data D1a to store data DKa of all stores, a record of the same specified date (indicating purchase date) and the same product is extracted from each store data D1a, D2a,..., DKa.

  For example, in the date data set in “Year / Month / Day” 611, two months before the aggregation date are first designated. Furthermore, the record group with the smallest value of “product code” 633 (code 1 in this example) is designated in the record group hit on that day. Then, one record specified by the date and the product code is extracted from each store data D1a, D2a,..., DKa. After extraction, the number of daily purchases set in each record is aggregated. Then, the value of “number of daily purchases” 662 of one of the extracted records (for example, one record extracted from the store data D1a) is updated with the total value. Finally, the record generated by this update is stored in the past information database as total shop data DTa.

  Similarly, records are sequentially generated for other products on that date. That is, two months before the aggregation date is designated, the code 2 next to the code 1 is designated, the aggregation operation is performed in the same manner, and the generated record is added to the aggregation store data DTa. After performing the totaling operation for all products on the first designated date, the next day is specified, and the totaling operation is performed in the same manner from code 1 in order, and the generated record is added to the total store data DTa. In addition, “000” is set in the “store code” 601 as the identification code of the total store.

  As a result, as shown in FIG. 6, in the aggregate store data DTa, the value of “daily purchase quantity” 662 is shown as an aggregate of the daily purchase quantity of all stores. Taking the record in the first row of the figure as an example, the purchase record of all stores of the product “Code 1” on the purchase date “January 16, 2014” is “801” in the value of “daily purchase quantity” 662. Is set.

  Next, store data D1b to store data DKb of the future information database 311 (schedule information 311A) will be described.

FIG. 7 is store data D1b of the store 1 shown as an example of the schedule information 311A.
Store data D1b of FIG. 7 includes “store code” 701, “date of month” 711, “day of the week” 712, “holiday” 713, “scheduled maximum temperature” 721, “scheduled minimum temperature” 722, “scheduled precipitation”. 723, “Line Code” 731, “Class Code” 732, “Product Code” 733, “Scheduled Standard Price” 741, “Scheduled Sale Price” 742, “Scheduled Discount Price” 743, “Scheduled Flyer Publication” 744, “Scheduled Display” Items such as “status” 751 are included.

“Store code” 701 is an item indicating a store identification code.
“Date” 711, “Day of the week” 712, and “Holiday” 713 are items indicating future “Date”, “Day of the week”, and “Holiday” for which the demand is predicted.

“Scheduled maximum temperature” 721, “Scheduled minimum temperature” 722, and “Scheduled precipitation” 723 are items indicating meteorological data on a future date for which demand prediction is performed.
The “line code” 731, “class code” 732, and “product code” 733 are the same as those described in the “line code” 631, “class code” 632, and “product code” 633 in the past information, respectively. The description here is omitted.

The “scheduled standard price” 741 is an item indicating the product price at the time of standard sales on a future date for which demand prediction is performed.
The “scheduled selling price” 742 is an item indicating a commodity price at the time of special sale on a future date for which demand prediction is performed.

The “scheduled discount amount” 743 is an item indicating a discount amount from the scheduled standard price. Here, an amount obtained by subtracting the planned selling price from the planned standard price is set.
“Planned flyer publication” 744 is an item indicating a sales promotion method on a future date for which demand prediction is performed.
The “scheduled display state” 751 is an item indicating the sales state of the product on a future date for which the demand prediction is performed.

Below each data item, a part of the schedule data of the store 1 is shown.
One record indicates schedule data of one product for each scheduled purchase date as in the configuration of the store data D1a of past information. In the example of the figure, as indicated by the data of “Date” 711 and the data of “Product Code” 733, first, records corresponding to the number of products indicating the scheduled data of “March 17, 2014” are shown. Subsequently, records corresponding to the number of products indicating the scheduled data of “March 18, 2014” on the next day are shown. After that, although omitted, data for several weeks or months ahead is included.

Next, the summary store data DTb of the future information database 311 (schedule information 311A) will be described.
The aggregate store data DTb is generated by copying any one of the schedule information of the store data D1b, D2b,..., DKb. For example, when the store data D1b is copied and generated, it is generated by changing the value of the “store code” 701 to “000”. It should be noted that since the regional data does not make sense in the aggregation store, all may be set to “Null”. Hereinafter, the regression coefficient of the regional data is set to “0” so that the regional data is not reflected in the prediction formula. Therefore, the regional data is not changed to “Null” and is left as it is.

Returning to FIG. 4, the continuation will be described.
The regression coefficient calculation unit 313 calculates a regression coefficient (including a constant term) indicating the degree to which past variable factor information contributes to the fluctuation of the objective variable. Multiple variables contribute to the fluctuation of the objective variable. Here, each regression coefficient of a plurality of variation factors is calculated for each product, and a calculation result (data including the value of each regression coefficient) is output for each product.

  In the present embodiment, the regression coefficient calculation unit 313 calculates a regression coefficient for each store. That is, the regression coefficients of the total store (store 0), store 1, store 2, ..., store K are calculated.

  Equation (1) is applied to both the calculation of the regression coefficient of the store 1 to the store K and the calculation of the regression coefficient of the total store (store 0). In the calculation of the regression coefficient of the store 1 to the store K, the regional data and the non-regional data are used as they are as purchase performance data. On the other hand, in the calculation of the regression coefficient of the aggregation store, the regional data does not make sense and must be excluded. In this embodiment, the regression coefficient of the variation factor corresponding to the regional data is set to “0” in advance. Thereby, also in calculation of a total shop, area data and non-area data are used as they are, and the same formula (1) as that for each shop is applied.

The calculation of the regression coefficient for each store and the calculation of the regression coefficient for the total store are as follows.
In the case of each store, the store data of one store is used, and the past daily number of purchases of the product n is substituted for the objective variable Yn in the equation (1). Further, the number of daily purchases of other products (substitute on the same day as that substituted for Yn) is substituted for the fluctuation factors Y1,..., YN. Further, past variation factor information is substituted for Xn1,..., XnM. Since there are M + N regression coefficients in equation (1), use M + N or more sets of data of past purchases and variable factor information, formulate M + N or more equations, and calculate the objective variable using the conventional regression coefficient calculation method. The regression coefficient of Yn is obtained. This operation is performed for the objective variables Y1,..., YN, and a regression coefficient is obtained for each product n. At this time, a known correction conditional expression (for example, a least square method) is introduced in order to improve the prediction accuracy.

The above operation is performed on stores 1 to K, and a regression coefficient for each store is obtained.
On the other hand, the total store uses the total store data, and substitutes the past daily number of purchases of the product n into the objective variable Yn of the equation (1). Further, the number of daily purchases of other products (substitute on the same day as that substituted for Yn) is substituted for the fluctuation factors Y1,..., YN. Further, past variation factor information is substituted into Xn1,..., XnM, and the regression coefficient of the variation factor corresponding to the regional data is set to “0”. Since there are M + N− (the number L of the variation factors of the regional data), the regression coefficient of the formula (1) is M + N−L or more by using M + N−L or more sets of past purchases and variation factor information data. And the regression coefficient of the objective variable Yn is obtained by a conventional regression coefficient calculation method. This operation is performed for the objective variables Y1,..., YN, and a regression coefficient is obtained for each product n. At this time, a known correction conditional expression (for example, a least square method) is introduced in order to improve the prediction accuracy.

By this operation, the regression coefficient of the aggregation store with the regression coefficient corresponding to the regional data being “0” is obtained.
Note that these regression coefficient calculation processes are performed in advance at night or the like.

The regression coefficient database (DB) 314 stores the regression coefficient for each store and each product calculated by the regression coefficient calculation unit 313.
Here, the regression coefficients for each of the products 1 to N of the store 0 (aggregate store), store 1, store 2,..., Store K are stored.

  In the forecast purchase number calculation unit (demand prediction means, store-specific demand prediction means) 315, the demand prediction calculation of each product for the total store or for individual stores (store 1, store 2,..., Store K). I do. Specifically, the following processing is performed for each store. First, various future fluctuation factor information stored in the schedule information 311A of the product n and the regression coefficient stored in the regression coefficient database 314 (the regression coefficient of the product n) are applied to the prediction formula (1). . This calculation procedure is performed for the products 1 to N, and simultaneous equations corresponding to the products 1 to N are established. From these equations, the predicted purchase numbers Y1,..., YN of all the corresponding products (product group) are calculated. The predicted number of purchases as a calculation result is output as data.

  In the present embodiment, this series of processing is first performed for the aggregation store. Subsequently, the series of processes is performed in the order of store 1, store 2,..., Store K, and the predicted number of purchases of each store including the total store is calculated.

  The forecast information 311B of the future information database 311 includes information such as forecast purchase data 326d, forecast inventory data 327d, and recommended order data 328d. The predicted purchase number for each store and each product output from the predicted purchase number calculation unit 315, the predicted sales amount calculated based on the predicted purchase number, the predicted profit amount, and the like are stored in the predicted purchase data 326d.

  The order / inventory number calculation unit 316 subtracts the predicted purchase number calculated by the predicted purchase number calculation unit 315 from the current inventory number, and predicts the estimated inventory number at a set time in the future (second predetermined period). Is calculated. Further, the order / inventory quantity calculation unit 316 increases the recommended order quantity based on the calculated predicted inventory quantity, for example, if the predicted inventory quantity is less than the minimum inventory quantity, and recommends if the predicted inventory quantity exceeds the minimum inventory quantity. The number of orders is reduced and finally processed into recommended order data according to a predetermined order unit and order lead time. The prediction information 311B stores the predicted inventory quantity (predicted inventory data 327d) calculated by the order / inventory quantity calculation unit 316 and the recommended order quantity (recommended order data 328d).

  The input reception unit 317 generates various screens of the demand prediction system (demand prediction setting screen, plan editing screen, notification confirmation screen, store-specific prediction screen, etc.) as result output means and store-specific result output means. Various screens are transmitted to the planning support system 330 and the order quantity determination support system 331. The plan edit screen is transmitted including demand forecast data, instruction means for displaying the store-specific forecast screen, an input unit for changing the value of the demand forecast data, and the like. In addition, a notification screen, a notification signal, and the like are transmitted as notification means for notifying that the forecast calculation for each store is completed in the predicted purchase number calculation unit 315. The headquarters special sales planning terminal 124 and the store ordering terminal 125, which are external devices, access the special sales planning support system 330 and the order quantity determination support system 331, and display various displays (display areas such as liquid crystal) as interfaces for simulation and the like. ).

  Further, the input receiving unit 317 uses the data (for example, product code, selling price, day of the week, holiday, temperature, precipitation, regional event, presence / absence of flyer publication) input from various screens as the schedule information 311A, and the future variable factors. Register as information. Further, when the input data is corrected from various screens, various future variable factor information stored in the schedule information 311A is updated to the corrected values. In addition, when there is registration or correction of future variable factor information, a demand forecast simulation execution command is output to the forecast purchase quantity calculation unit 315.

  Further, in the input reception unit 317, the calculated predicted purchase number, predicted stock number, and recommended order number are the target values of the predicted purchase number, the predicted stock number, and the recommended order number that are input from the store ordering terminal 125. When the target value is not satisfied, the forecast purchase quantity is recalculated by correcting various future fluctuation factor information stored in the schedule information 311A as the forecast condition, and the calculated forecast value (predicted purchase quantity) , Forecast inventory quantity, recommended order quantity)

FIG. 8 shows an example of a configuration diagram of the plan editing screen of the headquarters.
The plan edit screen 4G in FIG. 8 is a screen used for processing for determining the number of orders by displaying demand forecast data and editing data of products for sale plan.
The plan edit screen 4G shown in FIG. 5 is displayed by displaying a list of plans on the demand prediction setting screen (not shown) and selecting one plan.

  The plan edit screen 4G shown in FIG. 8 includes a plan display area 41, a product group selection area 42, a data display area 43, a data operation area 44, a store-specific prediction confirmation button 45, and various buttons. A navigation button area 46 is included.

  The plan display area 41 is provided as an area for displaying the “plan No.” 410, “plan name” 411, “start date” 412, and “end date” 413 of the plan selected on the demand prediction setting screen of the previous screen. Yes.

  The product group selection area 42 is provided as a selection area for displaying part or all of the product group to be planned in the data display area 43 by user selection. The product group selection area 42 includes a range designating section divided into “department” 420-1, “line” 420-2, and “class” 420-3. In the “department” 420-1, the range of the highest product classification category can be designated. In the “line” 420-2, the range of the product classification category immediately below the “department” 420-1 can be designated. In the “class” 420-3, the range of the lowest product classification category can be designated. Each range designating section is provided with a check box B1 for designating all of the displayed product category categories as a range.

  The data display area 43 is provided as an area for displaying demand prediction data for some or all products in the product group to be planned. The demand forecast data displayed here is a result of demand forecast calculation based on the aggregate store data DT. As display items of demand forecast data, “single item code” 430, “manufacturer” 431, “product name” 432, “standard” 433, “sales (purchase) date” 434, “special sale price” 435, “forecast sale (purchase)” "Number" 436, "Predicted sales (purchase) amount" 437, "Predicted profit amount" 438, "Actual sales (purchase) number" 439, "Actual sales (purchase) amount" 440, "Actual profit amount" 441 are provided. ing. In the data display area 43, by default, demand prediction data is displayed for all of the product group. The range of the product group displayed by the designation in the product group selection area 42 is limited.

The data operation area 44 is provided as an area for performing data editing such as correction of product data in the data display area 43. In the data operation area 44, an input box B2 for editing data and a correction button B3 are provided.
The input box B <b> 2 includes an edit box for each item of demand forecast data in the data display area 43.

  When one product is selected in the data display area 43, the data of this product is displayed in each input box B2 so that it can be edited.

  The correction button B3 is a declaration for correcting the record information of the product designated on the data display area 43 for each item. For example, one product is designated in the data display area 43, and the data of this product is displayed in each input box B2. Then, the data displayed in the predetermined input box B2 is rewritten, and the correction button B3 is pressed. As a result, the product data is corrected, and the corrected data is displayed in the data display area 43.

  When the data-corrected item corresponds to a variable factor, the demand forecast is recalculated, and the demand forecast data after the recalculation is displayed in the data display area 43. In this embodiment, in particular, the “sale price” corresponding to the variable factor is corrected. The “special sale price” may be corrected by batching the special sale price for the entire planning period of one product designated in the data display area 43 or only the special sale price on the designated sale date. In the following description, it is assumed that the special sales price for the entire project period is corrected in a lump.

  The store-specific prediction confirmation button 45 is an instruction button (instruction means) for displaying a result of a demand prediction simulation based on purchase result data for each store (store 1, store 2,..., Store K). . Since the demand prediction simulation is performed for each store, the displayed result of the demand prediction calculation reflects the regional data of the store.

  The navigation button area 46 is provided as an area for displaying screen operation buttons common to other screens. On the plan edit screen 4G, an end button B6, a print button B7, a return button B8, and a confirm button B9 are displayed so as to be operable. The end button B6 is a button for declaring that the demand prediction setting is ended. The print button B7 is a button for outputting a print command for the plan edit screen 4G to the printer. The return button B8 is a button for returning to the previous demand prediction setting screen. The confirmation button B9 is a button for confirming the setting contents of the data display area 43 and instructing an order.

<Processing of demand forecasting device>
FIG. 9 is a flowchart of processing relating to data collection and demand prediction simulation performed at night in the demand prediction apparatus.
This process is started once a day at a fixed time (nighttime), initializes the variable i = 0, and performs the following processes.

  First, data collection processing is performed (S1). In the data collection process in step S1, past information and schedule information are collected for each store via a communication network. As past information, past purchase record data accumulated in each store server 120, meteorological data of a weather system, and the like are collected. As schedule information, future weather forecast data of the weather system, calendar information, etc. are collected. It is assumed that information related to products in a special sale plan scheduled in the future is registered at any time from the special sales plan terminal 124 of the headquarters, etc., and has been registered up to a schedule several weeks ahead.

  Through the process of step S1, store data D1a to store data DKa consisting of past information are stored in the past information database 310, and store data D1b to store data DKb consisting of future data are stored as future information 311A in the future information database 311. .

  Next, aggregate store data DTa is generated (S2). In the process of step S2, the store data D1a to store data DKa stored in the past information database 310 are totaled to generate aggregate store data DTa as purchase record data of the aggregate store (store 0) that is regarded as one store. To do. By generating the aggregate store data DTa, it is possible to obtain a regression coefficient that reflects purchase record data of all stores when a simulation for demand prediction is performed.

Next, it is determined whether or not the value of the variable i is larger than K (S3).
If it is determined in step S3 that the value is K or less (No determination), a regression coefficient corresponding to the store data Dia is calculated (S4). In the process of step S4, since i = 0 in the first cycle, the regression coefficient corresponding to the store data D0a (that is, the total store data DTa) is calculated. In the process of calculating the regression coefficient corresponding to the aggregate store data, the variation factor of the regional data has no meaning. Therefore, in the prediction formula (1), the regression coefficient of the variation factor of the regional data is set to “0”, the past performance data is assigned to each variation factor, and the remaining regression coefficient is calculated.

After the process of step S4, the obtained regression coefficient is stored in the regression coefficient DB 314 (S5).
Next, the value of i is incremented (S6), and the process returns to the determination process in step S3.
In step S4, since i> 0 and the calculation target of the regression coefficient is the store data D1a, the store data D2a,. Becomes meaningful. In this case, in the prediction formula (1), all the regression coefficients are calculated by applying past performance data to each variable factor.

If it is determined in step S3 that it is greater than K (Yes determination), i = 0 is set (S7), and the process proceeds to the determination process in step S8.
In step S8, it is determined whether or not the value of the variable i is larger than K.
If it is determined in step S8 that the value is K or less (No determination), demand forecast data corresponding to the store i is demanded using the regression coefficient obtained based on the store data Dia and the schedule data of the store data Div. It calculates by prediction calculation (S9).

  Since i = 0 in the first round, the value of the regression coefficient corresponding to the store data D0a (that is, the total store data DTa) is substituted into the regression coefficient of the formula (1), and the store data D0b (total store data DTb) is expressed by the formula ( Substituting into the corresponding variable factors in 1), the estimated number of purchases of each product that will be the remaining unknown is obtained. Further, the forecast sales amount and the forecast profit amount are obtained based on the forecast purchase quantity, and the demand forecast data including them is configured.

After the process of step S9, the obtained demand forecast data is stored in the forecast purchase data 326d of the forecast information 311B (S10).
Next, the value of i is incremented (S11), and the process returns to the determination process in step S8.

  In step S9, after the second round, i> 0 and demand calculation data is calculated for each store 1, store 2,. In this case, the value of the regression coefficient corresponding to the store data Dia is substituted into the regression coefficient of the equation (1), the store data Dib is substituted into the corresponding variable factor of the equation (1), and the remaining unknowns for each product. Find the predicted purchase quantity. Further, the forecast sales amount and the forecast profit amount are obtained based on the forecast purchase quantity, and the demand forecast data including them is configured.

  Because the demand forecast data for the first round was calculated using regression data using purchase data for all stores, the data parameter is overwhelmingly large compared to the demand forecast for each store after the second round. If regional characteristics are ignored, highly reliable demand forecast data can be obtained.

Since the demand forecast data after the second round considers regional characteristics, the demand forecast data is highly reliable as demand forecast data for each store.
If it is determined in step S8 that the value is greater than K (Yes determination), this process ends.

  FIG. 10 is a flow diagram of a demand prediction simulation process performed between the headquarters special sales planning terminal operated by the buyer at the headquarters and the demand prediction device. It is assumed that the special sales planning support system 330 and the order quantity determination support system 331 are interposed as an interface between the head office special sales planning terminal 124 and the demand prediction device 31.

Broken line arrows shown in the figure indicate the flow of signals between the headquarters special sales planning terminal 124 and the demand prediction device 31.
First, in the headquarters special sales planning terminal 124, the buyer activates the demand prediction setting program and displays the demand prediction setting screen on the display screen (S20).

  Next, a plan list is displayed, and one plan is selected from them (S21). By this selection, an instruction signal for acquiring demand prediction data of one plan is transmitted from the headquarters special sales planning terminal 124 to the demand prediction device 31.

  The demand prediction device 31 determines whether an instruction signal has been received (S30). When the instruction signal is received from the headquarters special sales planning terminal 124, the determination process of step S30 is Yes, and the plan edit screen 4G is transmitted to the headquarters special sales planning terminal 124 (S31). The plan editing screen 4G includes an input unit that can input the correction value of the selling price and instruct the demand prediction device 31 to correct the value by this value. Moreover, the demand forecast data of one plan selected on the demand forecast setting screen are included.

  This demand forecast data is data calculated based on the total store data DTa stored as forecast purchase data 326d in the forecast information 311B of the future information database 311.

On the other hand, the head office special sales planning terminal 124 displays the plan editing screen 4G (S22).
Although not shown in FIG. 10, it is assumed that the buyer presses the store prediction confirmation button 45 on the plan editing screen 4G at this stage. In that case, the demand forecast data for each store stored as the forecast purchase data 326d in the forecast information 311B of the future information database 311 of the demand forecasting device 31 is included in the confirmation screen and supplied to the headquarters special sales planning terminal 124. .

  Now, when the buyer corrects the demand forecast data on the plan edit screen 4G, the headquarters special sales planning terminal 124 transmits a signal instructing re-simulation of demand forecast to the demand forecasting device 31 (S23). Specifically, if the sale price is to be corrected, one record is designated in the data display area 43 of the plan editing screen 4G, and the value of the sale price input box B23 in the input box B2 is corrected. Then, the correction button B3 is pressed, and a re-simulation request signal including the correction value of the sale price in the input box B23 is transmitted to the demand prediction device 31.

  Upon receiving the signal from the headquarters special sales planning terminal 124, the demand prediction device 31 determines whether or not the return button B8 has been operated based on the signal (S32) and determines whether or not the demand prediction data has been corrected (S33). ) Etc.

  In the case of Yes determination in step S32, it indicates that the return button B8 has been operated on the plan editing screen 4G, so the process returns to the determination process of step S30. In the case of No determination in step S32, the process proceeds to step S33.

  In step S33, it is determined whether the demand forecast data has been corrected. If there is a correction (Yes determination), the process proceeds to step S34.

In step S34, recalculation of demand prediction (re-simulation) for each product is performed based on the correction value.
At this time, the re-simulation is performed for the aggregation store. That is, the predicted purchase number of each product is calculated by the equation (1) using the regression coefficient of the aggregation store and the schedule information of the aggregation store.

In addition, the schedule information of each store shall be updated to the corrected value.
In addition, in the prediction information 311B of the future information database 311, the demand prediction data of the aggregation store is updated to the data after recalculation.

  After the processing in step S34, the recalculated demand forecast data is transmitted to the headquarters special sales planning terminal 124 (S35), and subsequently, the above correction values are used to re-simulate each store (S36). In step S36, for each store (store 1, store 2,..., Store K), the predicted number of purchases of each product is calculated using equation (1) using the store-specific regression coefficient and schedule information. As a result, demand forecast data for each store is obtained, and the demand forecast data for each store is updated to the data after recalculation in the forecast information 311B of the future information database 311.

When the re-simulation for each store in step S36 is completed, a notification signal is sent to inform the headquarters special sale planning terminal 124 (S37).
After the process of step S37, the process returns to the determination process of step S32.

On the other hand, the headquarters special sales planning terminal 124 receives the demand prediction data of the aggregate store transmitted as the re-simulation result of step S35 of the demand prediction device 31 after the processing of step S23, and displays the data display area 43 of the plan edit screen 4G. Information is updated (S24).
Thereafter, when the headquarters special sales planning terminal 124 receives the notification signal of step S37 of the demand prediction device 31, it displays in a pop-up that the demand prediction calculation for each store is completed (S25).

  When the buyer confirms that the demand prediction calculation for each store is completed by pop-up, the buyer requests the demand prediction data for each store from the demand prediction device 31 by pressing the store prediction confirmation button 45 (S26). .

  In the case of No determination in step S33, the demand prediction device 31 determines whether a request signal for demand prediction data for each store has been received (S38).

  In the determination in step S38, a Yes determination is made by the process in step S26 of the headquarters special sales planning terminal 124. In the case of Yes determination in the determination of step S38, a screen including demand prediction data for each store is transmitted to the headquarters special sales planning terminal 124 (S39), and the process returns to the determination process of step S32.

  In addition, the demand prediction apparatus 31 determines whether the end button B6 was pushed down in the case of No determination in the determination process of step S30 (S40). In the case of Yes determination that the end button B6 has been pressed in the determination process of step S40, this process ends. If the determination process in step S40 is No, the process returns to the determination process in step S30.

  If the determination in step S38 is No, it is determined whether or not the end button B6 has been pressed (S41). In the case of Yes determination that the end button B6 is pressed in the determination process of step S41, this process ends. In the case of No determination in the determination process of step S41, the process returns to the determination process of step S32.

  The headquarters special sales planning terminal 124 receives a screen including demand prediction data for each store as a result of the transmission processing in step S39 of the demand prediction device 31, and stores demand prediction data for each store in a window different from the plan editing screen 4G. Is displayed (S27).

When the buyer confirms the demand forecast data for each store and closes this window (S28), the buyer returns to the process of step S23 and accepts an operation on the plan edit screen 4G.
At the headquarters special sales planning terminal 124, the buyer inputs a correction value of the selling price, repeats the re-simulation, and finally determines the selling price. Then, the number of orders is determined based on the predicted purchase number of the demand prediction data.

  In the main process of the headquarters special sales planning terminal 124, when the return button B8 on the plan editing screen 4G is pressed by the buyer, a signal for instructing “return” is transmitted to the demand forecasting device 31 at an arbitrary timing. Returning to the process of S20, the demand prediction setting screen is displayed. When the end button B6 is pressed, a signal instructing termination is transmitted to the demand prediction device 31 at an arbitrary timing, and the process is terminated. The demand prediction device 31 detects that there is an operation of the return button B8 at the headquarters special sales planning terminal 124 in the determination of step S32. Further, it is detected in step S40 and step S41 that the end button B6 is operated at the headquarters special sales planning terminal 124.

  A series of re-simulations based on the correction of the demand forecast data described above will be described on the plan edit screen 4G transmitted as an interface to the headquarters special sales plan terminal 124 as follows.

  The plan editing screen 4G shown in FIG. 8 includes a plan name “Department 1 Week Special Sale”, a plan period “Monday 17 March 2014 to 23 March 2014 (Sun)”, and a product category “Department 1”. The existing demand forecast simulation result of the goods which belong to is shown. Therefore, this represents the state of the plan edit screen shown in step S22 of FIG.

  In the data display area 43, a part of the special sale products belonging to “Department 1” is shown. By operating the scroll bar Bar1, other products are displayed in the data display area 43 in the order of single product codes, and demand forecast data can be confirmed over all the special sales products belonging to “Department 1”.

  In FIG. 8, the record Rec01 in the first line of the product category “class 1” is indicated by a thick frame. This is used to indicate that the record Rec01 is designated.

  In the input box B2 of the data operation area 44, the demand forecast data of the commodity of the record Rec01 in the first line designated in the data display area 43 is displayed.

  In the plan editing screen 4G shown in FIG. 8, for example, the value “250 (yen)” shown in the input box B23 of “sale price” in the data displayed in the input box B2 is changed to “210 (yen)”. correct. Then, the correction button B3 is pressed (corresponding to step S23 in FIG. 10).

  By this correction operation, the processing of steps S34 and S35 of FIG. 10 is performed, and the data after re-simulation (demand forecast data of the aggregate store) is displayed in the data display area 43 (corresponding to step S24 of FIG. 10). .

FIG. 11 shows a display example of the plan edit screen after the re-simulation.
In the plan edit screen 11G of the figure, the data shown in the record Rec10 is data after the data shown in the record Rec01 of the plan edit screen 4G in FIG. 8 is updated.

  The corrected value “210” is indicated as the value of the “special sale price” 435 of the record Rec10. Further, values “800”, “168,000”, and “20,000” obtained by re-simulation are shown as values of “predicted sales number” 436, “predicted sales amount” 437, and “predicted profit amount” 438, respectively. .

  As described above, here, when the selling price is corrected, the selling price within the planning period of the product is collectively corrected. For this reason, the value of “Special Sale Price” 435 in the next line of the record Rec10 is also changed to “210”.

  Further, the data of the record Rec20 having the same sales date as the value “3/17” of the “sales date” 434 of the record Rec10 (data after the update of the record Rec02 in FIG. 8) is also corrected by the correction of the selling price. The value of will be affected. Therefore, the values of “predicted sales number” 436, “predicted sales amount” 437, and “predicted profit amount” 438 are also updated.

  Each data of the record Rec10 specified in the data display area 43 is displayed in the input box B2 of the data operation area 44 in FIG.

Thereafter, when the record is designated again and its sale price is corrected, a simulation using the correction value is performed according to the procedure described above, and the result is displayed.
In this example, a record is already specified on the display screen after re-simulation, but it may be unspecified after re-display and the value of each input box B2 may be shown empty. .

  Subsequently, steps S36 and S37 in FIG. 10 are performed in the demand prediction apparatus 31, and a confirmation screen indicating that the simulation for each store is completed is displayed in a pop-up on the headquarters special sales planning terminal 124 (corresponding to step S25 in FIG. 10). ).

FIG. 12 shows a state when a confirmation screen is displayed in a pop-up on the plan editing screen 11G.
In the confirmation screen 12G of FIG. 12, information indicating that the simulation for each store is completed is represented by character information. A confirmation button 120B is provided on the confirmation screen 12G. When the buyer presses the confirmation button 120B, the confirmation screen 12G is closed, and the plan editing screen 11G can be operated.

  Thereafter, when the store-specific prediction confirmation button 45 is pressed on the plan edit screen 11G, the demand forecast data for each store is transmitted from the demand prediction device 31, and the store is displayed in a separate window from the plan edit screen 11G at the head office special sales plan terminal 124. Another demand forecast data is displayed.

FIG. 13 is an example of a display screen (prediction screen for each store) of demand prediction data for each store.
The store-specific prediction screen 13G of FIG. 13 displays demand prediction data for each store for the record specified in the data display area 43 of the plan edit screen 11G.

This example shows the demand forecast data of each store for the single item code “Code 1” and the sales date “3/17”, assuming that the record Rec10 is designated in the data display area 43 of the plan edit screen 11G.
The store-specific prediction screen 13G includes a product information display area 130 indicating information of a record designated on the plan editing screen 11G, and demand prediction data for each store (store 1, store 2,..., Store K) of the product. And a prediction data display area 131 for each store indicating the above.

  The store-specific predicted data display area 131 includes items of “store code” 1310, “store name” 1311, “predicted sales quantity” 1312, “predicted sales amount” 1313, and “predicted profit amount” 1314. The data DL corresponding to is displayed.

Incidentally, the “store code” 1310 is an identification code for identifying each store. In FIG. 13, the identification codes are indicated by 001, 002,.
“Store name” 1311 is a name given to the store. In FIG. 13, the store names are indicated by store 1, store 2,.

In FIG. 13, data DL shows only data for two stores, and data for other stores is omitted.
Note that the scroll bar Bar <b> 2 is provided on the assumption that the data of all stores cannot be displayed in the store-specific predicted data display area 131. When the data of all stores cannot be displayed, the data of all stores is confirmed by operating the scroll bar Bar2.

Furthermore, a confirmation button 130B is provided at the bottom of the store-specific prediction screen 13G.
By pressing the confirmation button 130B, the store-specific prediction screen 13G is closed and the plan editing screen 11G can be operated.

  Thereafter, when another product or a record of other sales date of the same product is designated in the data display area 43 of the plan edit screen 11G and the store-specific prediction confirmation button 45 is pressed, the demand for each store corresponding to the record is displayed. Prediction data is generated as a store-specific prediction screen 13G.

  Further, when the sale price etc. is corrected in the data operation area 44 of the plan edit screen 11G, the re-simulation is performed again for the aggregation store, and the demand forecast data is displayed in the data display area 43. When calculation of the demand forecast for each store is completed, a message to that effect is displayed in a pop-up.

As described above, the buyer at the headquarters can determine the total order quantity of each product of all the stores with reference to the demand prediction data of the aggregate stores.
Further, the distribution of the order quantity to each store can be determined with reference to the demand forecast data for each store.

  The distribution of the order quantity to each store is performed by, for example, assigning to the input receiving unit 317 an apportioning means for apportioning the predicted purchase number included in the demand prediction data of the aggregate store by the predicted purchase number included in the demand prediction data for each store. This can be implemented.

The input receiving unit 317 transmits the order quantity for each store generated by the prorated means, including information indicating the distribution amount on the store prediction screen 13G.
The order quantity for each store is calculated as a distribution amount by multiplying the estimated number of purchases at the total store by the respective ratio for each store according to the ratio of the estimated purchase number at each store.

FIG. 14 is a display example of the store-specific prediction screen 14G including the distribution amount.
As shown in FIG. 14, a new item “distribution amount” 1401 is provided in the store-specific prediction data display area 131, and data indicating the distribution amount is displayed in the item.

Here, “35 (pieces)” for the store 1 and “33 (pieces)” for the store 2 are displayed as the results calculated by the prorated means.
The difference from the value of the “predicted sales number” 1312 is due to an error between the total predicted sales number of each store and the predicted sales number of the total store.

  In the present embodiment, an aspect of totaling purchase record data of all stores as total store data is shown, but this is not restrictive. You may total the purchase performance data of several stores in all the stores. In this case, the buyer can place a batch order for the plurality of stores.

  Further, in the present embodiment, it has been described that the input receiving unit 317 generates various screens (demand prediction setting screen, plan editing screen, notification confirmation screen, store-specific prediction screen, etc.), but this is not restrictive. . The input reception unit 317 transmits / receives demand forecast data, input data, instruction signals, etc. to / from the special sales planning support system 330 and the order quantity determination support system 331. A screen (a demand prediction setting screen, a plan editing screen, a confirmation screen for notification, a prediction screen for each store, etc.) may be generated. The functions of the headquarters special sales planning terminal 124 and the store ordering terminal 125 may be appropriately modified so as to generate various screens.

  As described above, the buyer at the headquarters can immediately obtain the result of the demand prediction calculation using the purchase record data of all the stores as a sample from the demand prediction device and analyze it on the screen. Moreover, the detailed result for every store can be acquired from a demand prediction apparatus during the analysis, and it can be confirmed on a screen.

  As a result, when the buyer at the headquarters places an order for products common to all stores, the demand forecast result of the aggregate store data is analyzed, and when placing an order for a unique product for each store, the demand forecast result for each store is It becomes possible to use properly such as analyzing.

  Various programs used in the demand prediction apparatus according to the present embodiment and the modification are files in an installable or executable format, such as a CD-ROM, a flexible disk (FD), a CD-R, a DVD (Digital Versatile Disk), and the like. The program may be provided by being recorded on a computer-readable recording medium and read into a flash ROM of a demand prediction device or the like.

  Further, the program may be provided by being stored on a computer connected to a network such as the Internet and downloaded via the network.

  As described above, the demand prediction device according to the present embodiment and the modification includes the first storage area that stores purchase record data of a plurality of stores, and the purchases of the plurality of stores that are stored in the first storage area. Generation means for generating purchase record data for one store by aggregating the record data, a second storage area for storing purchase record data for the one store generated by the generation means, and the second storage area Demand forecasting means for performing demand forecasting calculation based on the purchase record data of the one store stored in, and result output means for outputting the result of the demand forecasting calculation of the demand forecasting means.

Further, a demand forecast unit for each store that performs a demand forecast calculation for each store based on the purchase record data of the plurality of stores stored in the first storage area, and the demand forecast calculation for the demand forecast unit for each store. Instructing means for instructing output of the results, and per-store result output means for outputting the result of the demand prediction calculation of the demand prediction means for each store according to the instruction of the instructing means.
As a result, it is possible to immediately obtain the result of demand prediction for a plurality of stores.

  In the above embodiment and modification, although the structure about the demand prediction apparatus was demonstrated, embodiment and the modification are shown as an example and are not intending limiting the range of invention. The novel embodiments and modifications can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the spirit of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

DESCRIPTION OF SYMBOLS 1 Demand prediction system 2 Communication line 3 Computer 5 Communication line 31 Demand prediction apparatus 32 Data collection / conversion system 33 Planned data collection / conversion system 120 Store server 124 Headquarters special sales planning terminal 125 Store order terminal 310 Past information database 311 Future information database 311A Schedule information 311B Prediction information 312 Aggregation unit 313 Regression coefficient calculation unit 314 Regression coefficient database 315 Predictive purchase number calculation unit 317 Input reception unit 326d Predictive purchase data 330 Sale plan support system 331 Order quantity determination support system D1 to DK Store data DT Total store Data S Buyer G screen

JP 2012-150671 A

Claims (6)

A first storage area for storing purchase record data of a plurality of stores;
Generating means for generating purchase record data of one store by aggregating purchase record data of the plurality of stores stored in the first storage area;
A second storage area for storing purchase record data of the one store generated by the generation unit;
Demand prediction means for performing demand prediction calculation based on the purchase record data of the one store stored in the second storage area;
A result output means for outputting the result of the demand prediction calculation of the demand prediction means;
A demand prediction apparatus comprising: A store-by-store demand prediction means for performing a store-by-store demand prediction calculation based on purchase record data of the plurality of stores stored in the first storage area;
Instruction means for instructing output of the result of the demand prediction calculation of the store-specific demand prediction means;
A result output unit for each store that outputs the result of the demand prediction calculation of the demand prediction unit for each store according to an instruction from the instruction unit;
The demand prediction apparatus according to claim 1, further comprising: The purchase record data includes data depending on the region and data not depending on the region,
The demand prediction means performs the demand prediction calculation based on data not depending on the region of the purchase record data of the one store.
The demand forecasting device according to claim 1 or 2 characterized by things. Notification means for notifying that the demand forecast calculation for each store by the demand prediction means for each store is completed,
The demand prediction apparatus according to claim 2 or 3, further comprising: The demand forecasting means calculates a predicted purchase number by demand forecast calculation based on purchase record data of the one store stored in the second storage area,
The store-specific demand forecasting means calculates a predicted purchase number by demand forecast calculation for each store based on purchase record data of the plurality of stores stored in the first storage area,
Apportioning means for apportioning the predicted purchase number calculated by the demand prediction means by the predicted purchase number calculated by the store-specific demand prediction means;
A prorated result output means for outputting the result prorated by the prorated means;
The demand prediction apparatus according to claim 2, further comprising: A computer for controlling the demand prediction apparatus having a first storage area for storing purchase record data of a plurality of stores,
Generating means for generating purchase record data of one store by aggregating purchase record data of the plurality of stores stored in the first storage area;
A second storage area for storing purchase record data of the one store generated by the generation unit;
Demand prediction means for performing demand prediction calculation based on the purchase record data of the one store stored in the second storage area;
A result output means for outputting the result of the demand prediction calculation of the demand prediction means;
Program to function as.

Images