JP4132188B2 - Management simulation device and management decision support system using this management simulation device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention simulates the occurrence of product / service exchange behavior and business behavior in retail, restaurant, service, etc. in accordance with the behavior of customers, shop assistants, and managers in actual trade areas and stores. Management simulation equipment And this Management simulation equipment Related to management decision support system using
[0002]
[Prior art]
It is well known that the sales floor layout in stores, the product layout (shelf allocation) of merchandise shelves, the setting of special sales prices, the location conditions in the trade area, etc., in retail businesses, etc. have a significant effect on sales and profits. However, at present, most of the layout and other plans depend on the intuition and experience of business managers. In addition, a method for making a plan such as a layout as accurate as possible and a method for sales prediction have been proposed.
[0003]
For example, Japanese Patent Application Laid-Open No. 64-21571 discloses a technique relating to shelf allocation of products in a store. According to this publication, “If there is a [long] empty space on the shelf, and there are no [earning] and [long] products, and [selling] and [long] products, [selling] and Knowledge base only for store side business such as [1] Face [Increase] [Long] product, and if the sales amount is high and the gross profit margin is high, it will be regarded as a profitable product and the sales amount The product placement is automatically determined based on the characterization of the product based on the past sales performance, such as a high-sale product if the gross profit margin is at a high level.
[0004]
However, in an actual retail business, customers who buy other products flow to other stores in a product structure with only the best selling. In addition, because special products (selling products) have a small profit, there is a problem that the effects of special sales cannot be demonstrated without considering floor layouts and shelving arrangements that allow other products to be purchased in addition to special products. is there. For this reason, simply focusing on one business and making a decision does not lead to a result that the business person in charge achieves the final management indicators such as sales, number of customers, profits, and the like.
[0005]
Japanese Patent Application Laid-Open No. 6-35896 discloses a technique for predicting sales when opening a store in a commercial area. According to this publication, there are means to calculate the expected sales value of a store based on the number of customers who can visit in the trade area set around the new store opening location, the traffic volume of the road in contact with the new store opening location and the locality, etc. Calculate the sales forecast value for each factor directly from the past statistical data by means of calculating the store sales forecast value based on the location conditions of the store, and further influence the degree of influence between each factor Sales by adding the knowledge base in which judgment knowledge is stored in advance, the means for correcting each calculated sales forecast value based on the impact judgment knowledge in the knowledge base, and each sales forecast value after correction The sales are corrected by means for obtaining the predicted output value.
[0006]
However, in such a method that directly uses past statistical data, the same trend as past data is applied to the present or the future, so the state and environment of the store or trade area to be predicted are monotonous. Of course, if you make a change that is not, you cannot use past data. In particular, the conventional method based on the past statistical data of at least about one year against the severity of recent changes in the retail, restaurant, and service industries, or the frequency of opening and closing stores, has a prediction accuracy. Not only did it fall, but the prediction itself was meaningless.
[0007]
Furthermore, there is a method disclosed in Japanese Patent Application Laid-Open No. 64-88215 for simulating consumer trends in a store. This method includes consumer trend detection means for detecting consumer trends in a store in time series, branch position detection means for detecting a branch position where the walking direction changes from the detected consumer trend, and branch position Branch rate calculation means for calculating the proportion of consumers in each branch direction in the store is provided, and the branch point or branch of the flow line of the consumer at a specific position in the store which is statistical data from the consumer trend detection means The simulation is performed directly based on the rate. Similarly, in the simulation of the consumer's product selection behavior, the stand-by rate, contact rate, and purchase rate for each product, which are statistical data from the consumer trend detection means, are directly used.
[0008]
According to this simulation method, it is possible to perform a flow simulation of the consumer when the product arrangement in the store is changed, but actually, the branching rate data in the same product arrangement as the product arrangement to be changed to the past data is When there is no data, the simulation user inputs an arbitrary value for data such as the branching rate, and the accuracy of the simulation is extremely low. In addition, because the statistical data of the detected stoppage rate, contact rate, and purchase rate itself is related not only to the attribute of the target product but also to the product placement of the product in the store, the product placement was changed. Sometimes the same stop rate, contact rate, and purchase rate values cannot be used for the same product. Furthermore, since the stoppage rate, contact rate, and purchase rate of one product are not only related to the target product, but also to the product placement of all other products in the store, this simulation method is used. The accuracy of the sales simulation when the product arrangement is changed by using it is hardly expected.
[0009]
On the other hand, there is a method that it is sufficient if there is statistical data for many types of product arrangements, but it is costly to change product arrangements frequently in order to obtain data for many product arrangements in actual stores. It is also difficult because it causes unnecessary confusion for consumers. Even if it is possible, it is data at different time points in time series, and even when it is possible to collect data from many stores at the same time, the trade area and customer base are actually different in each store Therefore, it is difficult to use the data when the product arrangement is changed.
[0010]
[Problems to be solved by the invention]
Management management support systems for management, marketing, merchandising, and operations performed by business managers in the retail, restaurant, and service industries have traditionally limited functions for individual store openings and shelving plans. There are several examples, but comprehensive strategies and actions related to specific operations such as store opening plans, management strategies, floor layouts, assortments, shelf allocation, sales promotion, pricing, inventory setting, product ordering, etc. There is no support for decision making for planning and comprehensive evaluation.
[0011]
However, in the actual management of retail, restaurant, service, etc., these store opening plans, management strategies, floor layouts, assortments, shelving, sales promotion, pricing, inventory setting, product ordering, etc. interact. As a result, the final management indicators such as sales, number of customers, profits, etc. targeted by the person in charge are realized even if the decision is made by focusing on one business. It does not lead to such a result.
[0012]
In addition, the conventional simulation methods such as product placement and sales forecast do not simulate the purchase behavior of individual buyers / users, such as store selection, product / service selection, etc., but monthly, annual statistics, etc. When, where, who, what, why (like the product itself, price, store atmosphere, store clerk's preference, etc.) and how to purchase It is not a simulation to express. For this reason, it is difficult to accurately grasp the generation process of the customer's purchasing behavior, and it is difficult to provide effective decision support for a specific strategy / action combined.
[0013]
Furthermore, in these statistical methods, for example, well-known regression analysis has been generally used, but the same tendency as the past data is applied to the present or the future. The past data cannot be used when the conditions inside or outside the store or trade area change. In addition, even if past data is often applicable, for example, if a downward trend in sales appears, it will be recognized, but that alone will lead to new ideas such as strategies and actions that will help maintain and improve sales. It is not directly linked to the creation of In particular, for verification of a new idea such as opening a new type of store or changing the layout / operation in the store, there is no past performance data on the idea, so statistical processing itself is impossible.
[0014]
Therefore, claims 1 to 3 The described invention is a management simulation capable of faithfully simulating a product / service exchange behavior and a business behavior generation system with a real person's behavior and the sales state of the product by a relatively simple algorithm. apparatus I will provide a.
[0015]
Claims 4 The invention described in this document is intended to be used by business managers of retailers, restaurants, service businesses, etc. in the decision-making of management strategies, concrete plans for actions, and comprehensive evaluations. Providing a management decision support system that verifies ideas related to new strategies and actions, and recommends appropriate strategies and actions necessary for the business manager to reach the target business condition.
[0016]
[Means for Solving the Problems]
Invention of Claim 1 Is a quotient Product / service exchange behavior and business behavior generation system Subject to the simulation of Target space consisting of an array of multiple cells Passive elements in relation to the area elements that restrict the movement of passive elements that act on the behavior elements and the behavior elements that are passively actuated with respect to the behavior elements and the movement of the behavior elements and the placement of passive elements , Area element and action element arrangement state table, target element state table that is the target of action element among passive elements, action element state table, action element vs. target element response coefficient setting table, action element vs. purpose Target using the memory on which the behavior response count setting table and the behavior element scanning capability setting table are formed, and the data relating to the placement of the passive element, the zone element and the behavior element in the target space stored in each placement state table The action element in the target area is scanned from the action element scan angle and the scan distance stored in the action element scan capability setting table. Set all the passive elements within this scanning range as the target elements, and the attraction level of the action elements for these target elements, the target element state table, the action element vs. target element response coefficient setting table, the action element The purpose set in advance with respect to the type of the attraction element, with the element having the maximum attraction as the attraction element According to the action, the state table of the target element and the state table of the action element Change the data and change it to multiple Action Repeat row for element Equipped with a simulation computer Management simulation apparatus It is in.
[0017]
The invention according to claim 2 is the management simulation according to claim 1. apparatus In The target space is in the store, the behavior element is a customer agent, the target behavior is purchase behavior, accounting behavior, and exit behavior, and the simulation computer uses the target behavior preset for the type of attraction element to purchase When it is an action, the number of purchased items is reduced from the state table of the passive element corresponding to the attracting element, and the code of the purchased item and the number of purchases are stored in the state table of the action element corresponding to the customer agent. is there.
[0018]
The invention according to claim 3 is the claim 1 Management simulation described apparatus In The target space is in the store, the behavior element is a customer agent, the target behavior is purchase behavior, accounting behavior and exit behavior, the memory includes an accounting unit status table and a store accounting business status table, and the simulation computer is When the target behavior set in advance for the type of attracting element is accounting behavior, the total number of products purchased by the customer agent is extracted from the status table of behavior elements corresponding to the customer agent, and the accounting unit status is calculated from the total number of products. The number of products that can be accounted for per unit time stored in the table is repeatedly subtracted until the total number of products becomes zero, and if the total number of products becomes zero, sales data is stored in the store accounting business status table In what is there.
[0024]
Claim 4 The described invention The management simulation apparatus according to claim 1, actual data collecting means for collecting various state values necessary for simulation from a real product / service giving / receiving behavior and business behavior generating system, and various state values collected by the real data collecting means A real state value data table for storing At the start of simulation by simulation means The various state values stored in the real state value data table are stored in each of the arrangement state table of the passive element, the zone element, and the action element, the state table of the target element, and the state table of the action element. Initial state value As Means to secure and at the end of simulation by simulation means In the action element state table Final state The value Means to secure, Means for displaying final status values Is a management decision support system with
[0027]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described with reference to the drawings.
(First embodiment)
FIG. 1 shows a device for performing a simulation of a product / service giving / receiving behavior and a business behavior generating system using a computer. This device includes a simulation computer 1 for executing a simulation of a product / service giving / receiving behavior and a business behavior generating system. An input device 2 such as a keyboard and mouse for inputting various instructions and simulation conditions and simulation data to the simulation computer 1; a graphic display for displaying simulation results and operation screens; and an output device 3 such as a printer. It is constituted by.
[0028]
The simulation computer 1 includes a CPU (central processing unit) 4, a main memory 5 that stores an operation program, a simulation program, simulation data, and the like of the CPU 4 in a storage area, and a magnetic that stores the programs, data, simulation results, and the like. A storage device 6 such as a disk is provided, and a simulation program for performing data setting and data configuration is executed.
In the main memory 5, as shown in FIG. 2, a program storage unit 7a for storing various programs is formed, and an element arrangement state table 7b, target element state table 7c, and action element pairs in which various data are set. Target element response coefficient setting table 7d, action element state table 7e, action element target action response coefficient setting table 7f, work time division state table 7g, time increment setting table 7h, action element attribute state master table 7i, stationary work state Table 7j, history file 7k, action element scanning capability setting table 7m, target element scanning result data table 7n, action limit value setting table 7p, action distance limit value setting table 7q, customer agent state table 7r, accounting unit state table 7s, The store accounting business status table 7t etc. is formed Going on.
[0029]
FIG. 3 shows an example of cell division of a space (target space) to be simulated corresponding to a realistic store space. The two-dimensional target space 8 is expressed by an array of a plurality of square cells 9. ing. Although a hexagonal cell can be used as the cell, the square cell is better suited to the actual layout and is desirable for expressing the interior of the building and the like from the characteristics of the actual layout.
[0030]
Each cell 9 corresponds to position data (position coordinates) composed of a combination of two numerical values such as {i, j}, where i and j are arbitrary integers, for example, for the cell of interest 9c. . With this position data, all the cells in the target space 8 can be identified. Therefore, the positional relationship and the adjacent (joined) relationship of each cell 9 in the target space 8 can be expressed as follows using eight cells adjacent to the target cell 9c.
[0031]
The position data of the cell 9w on the left is {i-1, j}
The position data of the cell 9e on the right is {i + 1, j}
The position data of the upper cell 9n is {i, j-1}
The position data of the lower cell 9s is {i, j + 1}
The position data of the upper left cell 9nw is {i−1, j−1}.
The position data of the cell 9ne on the upper right is {i + 1, j-1}
The position data of the cell 9sw on the lower left is {i-1, j + 1}
The position data of the cell 9se at the lower right is {i + 1, j + 1}
In this state, the target space 8 merely shows a spatial expansion. In order to correspond to a realistic store, a plurality of elements (target elements) to be simulated are set in the target space. 8 should be placed on top. This combination of the target element and the target space is called a target region to be simulated.
[0032]
FIG. 4 shows an example of the target area 11 corresponding to a realistic in-store layout. The area element (on the space 8 in which no target element shown in FIG. Target elements called areas are appropriately arranged. That is, a passage area 12 that expresses a passage where customers and employees can freely move, a boundary area 13 that expresses walls, windows, etc. at boundaries that restrict movement and visibility of customers and employees, and products and sales promotion A shelf area 14 representing a table on which fixtures, billing machines, etc. on which in-store advertisements (POP) are arranged is arranged.
[0033]
In the area element, target elements called passive elements (units) are appropriately arranged. That is, as a passive element, there is an accounting unit 15 that represents an accounting machine (POS register) and related equipment, and the like is arranged in the passage area 12 or the shelf area 14. Further, as a passive element, there is an entrance / exit unit 16 that expresses an entrance / exit, and this unit is arranged in an area element between the inside and outside of the store.
The target area 11 corresponding to a realistic in-store layout can be generated by such area elements and passive elements.
[0034]
FIG. 5 shows an example of the target area 21 corresponding to the realistic sales floor configuration. On the target area 11 corresponding to the realistic in-store layout shown in FIG. A sales unit 22 representing an in-store advertisement (POP) or the like is mainly arranged on the shelf area 14 as a passive element. This makes it possible to represent a sales floor in a store where there are no customers or employees.
[0035]
In FIG. 5, a plurality of sales units 22 are arranged in all cells on all shelf areas 14, but for the sake of easy understanding, most of the sales units 22 on the shelf area 14 and the square cells 9 are omitted. Instead, a specific product group name represented by each sales unit 22 is written in the shelf area 14. In addition, products that represent food and beverage corners, toilets, parking lots, service counters, employee entrances, food processing rooms, warehouses, offices, employee break rooms, etc., in line with realistic retail stores. It is also possible to set target elements related to services and business.
[0036]
FIG. 6 shows an example of the target area 31 corresponding to the situation of product / service exchange and business occurrence in a realistic store. The target area 21 corresponding to the actual sales floor structure shown in FIG. In addition, an action element called an agent is added. As this behavior element, there is a customer agent 32 that represents a customer in the store (one person and a group performing the same behavior), and can move on the passage area 12 and the entrance / exit unit 16. In addition, there is an employee agent 33 representing an employee (one person and a group performing the same action) in the store, and can move on the aisle area 12 and the entrance / exit unit 16.
[0037]
This FIG. 6 shows the situation of the product / service exchange behavior and the business behavior generation system in the store, and realistic behaviors such as the purchase of the customer's product and the replenishment of the employee's product in the store. Is represented (simulated) by the behavior of the customer agent 32 and the employee agent 33 in the target area 31 in accordance with a simulation algorithm described later, which is the main part of the simulation of the present invention.
[0038]
The method of setting and arranging the target space and target elements is excellent in expressing the realistic situation in the store as faithfully as possible. However, in reality, the simulation algorithm is easy to create, the calculation capability of the simulation computer 1, and data In consideration of the balance with the ease of simulation such as storage capacity, it is desirable not to set up very finely at the expense of some fidelity as an actual model.
[0039]
Hereinafter, an example of setting the target area will be described.
In the cell setting of the target space, it is necessary to adopt a cell having the same shape and size for one target space in the store in order to easily create a simulation algorithm. In addition, the size of the square cell 9 in the target space includes a method of determining based on a distance corresponding to an average step of a person, a method of determining from an average size of furniture, etc. In order to express the above relatively easily, the cell size was set from the average reference dimensions of the fixture. Specifically, with 300 mm as a reference value, for example, the length of one side of the square cell 9 is made to correspond to a value of about 150 mm, about 300 mm, about 600 mm, or the like. Among these, a square cell with a side of about 300 mm (± 100 mm) is almost equal to the minimum floor area occupied by an average adult, and is convenient for algorithm creation. Model fidelity and ease of simulation The balance is excellent.
[0040]
In FIG. 6, one side is drawn as a square cell of about 600 mm so that the square cell 9 can be easily seen. In a store with a relatively large store area and a low congestion, the number of cells can be reduced and the simulation calculation time can be shortened by making each side a square cell of about 600 mm.
The basic setting of the target element is classified into the following three types according to the difference in function.
[0041]
As for the area element, only one area element can be arranged on one cell. The function of this area element is to limit the spatial behavior of the behavioral element and passive element, such as movement of the behavioral element, restriction of spatial scanning described later, placement of passive elements and restriction of passive direction described later. The direction that can be restricted is set. A plurality of passive elements can be arranged in area elements on one cell that can be arranged. The function of the passive element acts passively on the behavior element and induces (activates or suppresses) the behavior of the behavior element, and sets the direction in which the action can be performed. The action element can be located at a plurality of positions (entrance / appearance) in the area element and the passive element on one cell that can enter and exit. The function of this behavioral element interacts with the behavioral element and with the passive and area elements, and sets the direction in which the movement and action can be performed. The actual movement and action direction depends on the simulation algorithm. decide.
[0042]
If the above basic settings are satisfied, target elements can be set as appropriate in accordance with the actual store characteristics, and a simulation algorithm described later can be applied as long as the basic settings are followed. In addition, if the behavioral element is limited to one position (entrance / appearance) within the zone element and the passive element, the model's fidelity to this limitation occupies a square cell for an average adult. This can be accomplished relatively easily in terms of being approximately equal to the minimum floor area.
[0043]
Data relating to the arrangement of these target elements in the target space is recorded in the element arrangement state table 7b shown in FIG. It is also stored in the storage device 6. In the element arrangement state table 7b, an area element arrangement state table 35, a passive element arrangement state table 36, an action element arrangement state table 37, and the like are set.
[0044]
The area element arrangement state table 35 is set with respect to the position data of all cells included in the target area 31, and the type of area element arranged on the cell, the element code of the area element, the area element The element name etc. are stored. This element code is associated with all the target elements on the target area 31 on a one-to-one basis, whereby individual identification of the target elements can be performed.
[0045]
The element code of the area element is divided into, for example, three numbers by a point “.”. For example, in the element code “11.0001.0001”, the first number “11” is among the types of area elements. Indicating the boundary area, the intermediate numeral group “0001” indicates “wall” in the type of boundary area, the last numeral group “0001” indicates the first wall “wall 1”, and the element name “ Corresponds to wall 1 ”.
[0046]
Similarly, element codes are set for the other area elements, and the leading numeral group “12” indicates the passage area among the area element types, and the intermediate numeral group “0001” indicates the passage area type. The last digit group “0001” represents the first passage “passage 1” and corresponds to the element name “passage 1”. Similarly, the first numeral group “13” indicates a shelf area among the types of area elements, the intermediate numeral group “0001” indicates “shelf” among the types of shelf areas, and the last numeral group “ “0001” indicates the first shelf “shelf 1” and corresponds to the element name “shelf 1”.
[0047]
The passive element arrangement state table 36 is set for the element codes of all the area elements included in the target area 31 and including the passive elements, and the passive elements for all the passive elements arranged in the area element. Type, the element code of the passive element, etc. are stored.
[0048]
The element code of the passive element is divided into, for example, three number groups by a point “.”. For example, in the element code “23.0001.0001”, the first number group “23” is among the types of passive elements. Indicates the sales unit, the middle number group “0001” indicates the product group “soy sauce” in the type of sales unit, and the last number group “0001” indicates the individual product “a company” in the product group “soy sauce”. Soy sauce 1L (liter) ", corresponding to the element name" a company soy sauce 1L ". Similarly, the last number group “0002” corresponds to the individual product “b company soy sauce 1 L (liter)” in the product group “soy sauce”.
[0049]
Similarly, element codes are set for other passive elements, and the leading numeral group “21” indicates the entrance / exit unit among the passive element types, and the intermediate numeral group “0001” indicates the type of the entrance unit. Indicates the “front entrance”, and the last numerical group “0001” indicates the individual entrance unit “front entrance 1” in the “front entrance”, and corresponds to the element name “front entrance 1”. Similarly, the first numeral group “22” indicates the accounting unit among the passive element types, the intermediate numeral group “0001” indicates “accounting machine” among the accounting unit types, and the last numeral group. “0001” indicates an individual accounting unit “accounting machine 1” in “accounting machine” and corresponds to the element name “accounting machine 1”.
[0050]
In the passive element arrangement state table 36, the line for the area element in which the accounting unit is arranged and the line for the area element in which the entrance / exit unit is arranged are combined with the area element arrangement state table 35, and the in-store layout state table and It is called and used when generating the target area 11 corresponding to the in-store layout shown in FIG. Further, in the passive element arrangement state table 36, a table composed of rows for area elements in which sales units are arranged is called a sales floor structure state table, and the target area 21 corresponding to the sales floor structure shown in FIG. Used when generating.
[0051]
The behavior element arrangement state table 37 is set for the element codes of all the zone elements included in the target region 31 and including the behavior element, and the behavior element type for the behavior elements located in the zone element, The element code of the action element is stored.
[0052]
For example, in the element code “31.0001.0001”, the first numeral group “31” is the type of the action element. In the customer agent type, the middle number group “0001” indicates “male” in the customer agent type, and the last number group “0001” is “male” in the customer agent type. ”And corresponds to the element name“ A-like ”. Similarly, the middle number group “0002” indicates “female” among the types of customer agents, and the last number group “0002” corresponds to each customer “B” among the customer agents of “female”. is doing.
[0053]
Similarly, element codes are set for the other passive elements, and the first numeral group “32” indicates the employee agent among the action element types, and the intermediate numeral group “0011” indicates the employee agent type. In the figure, “male employee” is shown, and the last number group “0001” corresponds to the individual employee “employee D” in the “male employee”. In the same manner, the intermediate number group “0022” indicates “female part-time employee” among the types of employee agents, and the last number group “0002” indicates the individual number in “female part-time employee”. Corresponds to employee "Part E".
[0054]
Using such a table, a target area corresponding to a realistic store is constructed as a simulation model. The realistic store at this time does not mean only an actual store, but also includes an ideal store and a hypothetical store.
Next, an example of an in-store simulation algorithm will be described.
[0055]
FIG. 8 shows an algorithm for the entire in-store simulation, which is an algorithm for processing two types of behavior elements in the target area 31 in the store, that is, behaviors of the customer agent 32 and the employee agent 33.
First, in step S1, a simulation is initially set. Regarding the setting of the target area, an element arrangement state table preset up to the level of the target area 211 corresponding to the sales floor structure is called from the storage device 6 and stored in the main memory 5.
[0056]
Further, settings necessary for execution of the simulation, such as setting of a simulation period (start date / time, end date / time), are input from the input device 2 or are called from the storage device 6. In a realistic store, since a customer is affected by the purchase of goods from the season, day of the week, etc., the date is indispensable in setting the simulation period.
[0057]
The next loop L1 is a behavior processing loop within a behavior element simulation period, and repeatedly performs behavior processing from the simulation start date and time to the simulation end date and time.
[0058]
In step S2 in the loop L1, business time divisions such as a store opening preparation time, a business opening time, an end preparation time, and a non-working time corresponding to a realistic store are given to the simulation time. This refers to the business time division for the date and time preset in the business time division status table 7g.
[0059]
In the next step S3, in the behavior processing loop L1 within the behavior element simulation period, an increment of time, that is, a time step is given to the work time section. This refers to the increment value preset in the time increment setting table 7h. For example, the simulation can be accelerated with 1 time step being 1 second during business hours and 1 time step being 1 hour outside business hours such as regular holidays. This one time step is a unit time in the action processing loop L2 within the unit time of the next action element. This time step is used, for example, to set the number of cells in which the behavior element per time step can be moved when setting the movement speed of the behavior element, and is necessary to make the behavior of the behavior element realistic. Therefore, it is called a simulation target time, and the above-described target space and target elements and the target time are collectively referred to as a simulation target region.
[0060]
In the next step S4, processing for entering the behavior element into the store (in the target area) is performed, and the customer agent and the employee agent are processed. First, for a customer agent, a customer who enters the store at each time is determined based on a preset entry rule for the customer agent, for example, statistical data or a business area simulation described later. That is, one or a plurality of behavior element codes (customer codes) corresponding to customer agents entering the store are extracted from the behavior element attribute state master table 7i in which customer agents are registered in advance.
[0061]
In the behavior element attribute state master table 7i, the customer's age, sex, address, purchase history, etc. are recorded for the behavior element code. At present, this kind of data can be obtained relatively easily by a customer service card of the store or the company to which the store belongs, which has various privileges such as discounts.
[0062]
The store entry rule for the employee agent determines the employee who enters the store at each time as in the case of the customer agent, based on the employee assignment plan (work plan) set in advance. That is, one or a plurality of behavior element codes (employee codes) corresponding to employee agents who enter the store are extracted from the behavior element attribute state master table 7i in which employee agents are registered in advance.
[0063]
For example, a random number or the like is assigned to the action elements entering the store. Specifically, the action element code extracted is associated with the process number. As a result, the behavior element customer is subsequently processed until it leaves the store. The initial position of the customer agent is in the entrance / exit unit 16 as shown by the customer agent 32a entering the store in FIG. For the employee agent, the same entrance / exit unit is set as the initial position. These store entry rules also include a time when the store does not enter. At this time, the action element code is not extracted, and the process proceeds to the next loop L2.
[0064]
The loop L2 is a behavior processing loop within the unit time of the behavior element, and is a loop that is repeated until the behavior processing is completed for all the behavior elements in the store. In this loop, processing is performed in the order of processing numbers attached to the action elements. If there is no action element in the store, this loop is immediately exited and the process proceeds to step S8.
[0065]
In step S5 in the loop L2, each action process is branched according to the type of action element.
In step S6, the behavior processing of the customer agent within the unit time is performed.
In step S7, the employee agent performs action processing within a unit time.
When the action process of the loop L2 is thus completed, the next step S8 is performed.
[0066]
In step S8, a stationary operation that is necessary in the case of a retail store operation but does not require a simulation of the behavior of the employee agent in the store is performed. Here, processing based on the stationary work state table 7j in which the contents of the work are described is performed. Stationary work includes, for example, business related to sales promotion such as distribution of flyers to the business area to which the store belongs, payment of salary, creation of staffing plan, business related to accounting in the store, business related to store management, and the like. These operations are related to the behavior of the customer agent regarding the purchase of goods in the store and the behavior of the employee agent regarding the business. It is also necessary to calculate the profit and loss of the entire store.
[0067]
In the next step S9, processing of hh = hh + hhu, mm = mm + mmu, and ss = ss + ssu is performed, and the time is updated according to the time increment.
In the next step S10, the date is updated. At this time, the contents of the history file 7 k on the main memory 5 are stored in a storage file in the storage device 6. Thus, the main memory 5 is released so that a new data storage area can be secured in the main memory 5.
The above is the content of the behavior processing loop L1 within the behavior element simulation period, and the in-store simulation ends after the loop ends. The display of the simulation status is appropriately output by an output device 3 such as a graphic display or a printer in each process.
[0068]
The algorithm for the entire in-store simulation shown in FIG. 8 expresses that the behavior processing of the behavior element is performed according to a time step corresponding to a realistic time. This is because, for a certain period of time, such as sales forecasting, hot selling product analysis, and related purchasing product analysis by linear statistical analysis, such as multiple regression analysis, which has been often used in the retail industry. This is an essential difference from the one that calculates the total value.
[0069]
FIG. 9 shows an algorithm for the behavior processing of the customer agent in step S6 of FIG.
In step S11, the process first branches to step S12 of the next search action or step S13 of the accounting action according to the Mode variable that the customer agent has the action to be taken by the customer agent. The initial value of the customer agent 32a entering the store is “Mode = search”, and the search action step S12 is selected.
[0070]
As described above, the reason for selecting the search behavior and the accounting behavior by adopting the Mode variable is that once the accounting behavior is entered, the processing of the accounting behavior is not completed in one time step, and several time steps are required. Because.
[0071]
In step S12, it simulates the search behavior of the customer agent in the store, scanning the store, searching for the target, and moving and acting to search for and purchase a real customer product (put in the shopping basket). Selection of purchase behavior, accounting behavior at the end of shopping, exit from the store, etc. correspond.
[0072]
In step S13, the accounting behavior of causing the employee agent to account for the product purchased by the customer agent is simulated.
In step S14, a process of storing a data table necessary for the simulation among the various data tables in the history file 7k with the date and time is performed. This is because the contents of the data table change when the update process of the next step S15 is performed, so that the data table necessary for the simulation is stored in the history file 7k before that.
[0073]
In step S15, since the behavior element has performed some behavior, it is an update process for changing the behavior of the data table belonging to the behavior element. The contents of the data table include various status tables and setting tables.
Thus, the behavior processing for one customer agent within the unit time is completed, and the processing returns to the behavior processing loop L2 within the unit time in FIG. If there is an unprocessed behavior element, the behavior processing is performed.
[0074]
FIG. 10 shows an example of a search behavior algorithm, which is the most characteristic part of the present invention. Here, the search behavior in the customer agent behavior processing shown in step S12 of FIG. 9 will be described as an example.
First, in step S21, the customer agent scans the target area and performs a spatial scan to visually recognize the target passive element based on the scanning condition. And it is visually recognized as the scanning result. Receiving The moving elements are stored in the target element scanning result data table 7n.
[0075]
As a scanning condition, it is determined that the scanning method is visual scanning, the target element is a passive element, and the obstacle is a visual obstacle, and the scanning range of the corresponding customer agent set in advance from the behavior element scanning ability setting table 7m. The scanning angle and scanning distance are extracted and set.
This spatial scanning corresponds to an action in which a real customer looks inside a store and visually recognizes a product or the like. Specifically, as shown in FIG. 11, attention is first focused on the customer agent 41 that performs action processing. In the setting of the scanning range, it is assumed that the scanning angle is 180 ° centering on the front direction 42 of the customer agent 41 that performs the behavior processing, and the scanning distance has a value that can be viewed over the entire store. At this time, a blind spot occurs due to a visual obstacle 43 such as a shelf area, and the target area 31 is divided into a visible range 44 and an invisible range 45.
[0076]
Then, all the passive elements in the visible range 44 are extracted as being visible. At this time, all passive elements are extracted even when there are a plurality of passive elements in one cell. The element code of the extracted passive element and the distance, direction, etc. from the customer agent 41 that performs the action processing to the passive element are stored in the target element scanning result data table 7n.
[0077]
Next, in step S22, the degree of attraction for the target element of the customer agent 41 is calculated using all elements stored in the target element scanning result data table 7n as the degree of attraction calculation target.
As an example of the calculation method, the above-described target element state table 7c shown in FIG. 13, the above-described action element vs. target element response coefficient setting table 7d shown in FIG. 14, the above-described action element state table 7e shown in FIG. A description will be given of a method of calculating using the above-described behavioral element target behavior response coefficient setting table 7f.
[0078]
First, target element state table 7c For example, when the target element is a sales unit that represents a product or the like, the sales unit state table is used. This table 7c Stores the value of the product state and the sales state indicated by the sales unit for the element code of the target element (sales unit). For example, values such as price strength (%), discount rate (%), POP strength (%), and freshness (%) are stored. The price strength is a value obtained by dividing the price of the cheapest product among similar product groups to which the product belongs by the price of the corresponding product. The discount rate is a percentage of discount from the list price. The strength of POP (sales promotion advertisement in a store) is a value obtained by dividing the area of the product POP by the area of the largest POP in the similar product group to which the product belongs. Freshness is the reciprocal of the number of days elapsed from the date of production to the present. These values are set as the stimulus value Vi. The target element state table 7c is a value that can be obtained from an actual store.
[0079]
The action element versus target element response coefficient setting table 7d is, for example, a customer agent versus sales unit response coefficient table. This table 7d is set for each behavioral element code of the customer agent. For each passive element (sales unit) code, the state of the product indicated by the sales unit and the degree of response of the customer agent to the sales state are stored. Yes. For example, a price strength response coefficient, a discount rate response coefficient, a POP strength response coefficient, a freshness response coefficient, and the like are stored. The price strength response coefficient is a coefficient indicating the degree of response to the price strength, the discount rate response coefficient is a coefficient indicating the degree of response to the discount rate, and the POP strength response coefficient is the POP strength. It is a coefficient indicating the degree of response, and the freshness response coefficient is a coefficient indicating the degree of response to freshness. These values are defined as a reaction coefficient Ci. The action element versus target element response coefficient setting table 7d is a value that is difficult to obtain from an actual customer, and an estimated value is used here.
[0080]
From the above stimulus value Vi and response coefficient Ci, the attraction m is
m = Vi · Ci
Is a relatively simple method. There is also a method using a known sigmoid function or the like using two kinds of reaction coefficients in consideration of fidelity to reality.
[0081]
In this state, the search action targeting the sales unit of the customer agent does not end, and therefore, the calculated degree of attraction m is further suppressed here.
The behavior element state table 7e is, for example, a customer agent state table when the behavior element currently being processed is a customer agent. This table 7e is set for each behavioral element code of the customer agent, and stores values of various states of the customer agent at each time. For example, values such as the total number of purchased items, the total amount of purchased items, and the elapsed time since entering the store are stored. These values are referred to as stimulus values Vsi. This behavior element state table 7e is a value that can be obtained from an actual store.
[0082]
The behavioral factor response coefficient setting table 7f for behavioral elements is, for example, a customer agent purchasing behavior coefficient setting table. This table 7f is set for each purpose behavior (purchase behavior) of the customer agent, and stores the degree of response of the customer agent to various behaviors of the customer agent for each behavior element (customer agent). For example, the degree of reaction to the purchase behavior, the total purchase product amount, the purchase product total price, the elapsed time since entering the store, etc. is set as the reaction coefficient. These values are defined as a reaction coefficient Csi. This target behavioral response coefficient setting table 7f for behavioral elements is a value that is difficult to obtain from an actual customer, and an estimated value is used here.
[0083]
The attraction m is suppressed by the above stimulation value Vsi and the response coefficient Csi,
m = m− Vsi · Csi
As a result, it is a relatively simple method to finally obtain the attractiveness m. There is also a method using a known sigmoid function or the like using two types of reaction coefficients in consideration of fidelity to reality.
[0084]
As another suppression method, when values such as the total number of purchased products, the total amount of purchased products, and the elapsed time since entering the store exceed a set limit, the values of the action factor against target element response coefficient setting table 7f are the total number of purchased products, There is also a method of performing a suppression process that rapidly decreases in inverse proportion to values such as the total amount of purchased merchandise and the time elapsed since entering the store, by updating the data table in step S15 in the customer agent behavior processing algorithm of FIG.
Similarly, the degree of attraction is calculated for the accounting unit 15 and the entrance / exit unit 16 from the action element, the passive element, the state table for the target action and the response coefficient setting table as in FIGS. .
[0085]
In this way, the degree of attraction is calculated for all passive elements within the visible range, and the maximum value among them is taken as the maximum degree of attraction Mmax. An element having the maximum degree of attraction Mmax is set as an attraction element. Furthermore, the target action corresponding to the kind of attracting element is selected. That is, the target behavior in the case of a sales unit is a purchase behavior, the target behavior in the case of an accounting unit is accounting behavior, and the target behavior in the case of an entrance / exit unit is an exit behavior. In the case shown in FIG. 11, since the attraction element 46 is a sales unit, the target action is a purchase action.
[0086]
At this time, if the value of the total number of purchased items, the total amount of purchased items, the elapsed time since entering the store, etc. increases due to the suppression process, the degree of attraction for the sales unit (purchasing behavior) will decrease, and the relative accounting unit The degree of attraction for the gateway unit is increased, and the accounting unit is selected with the element having the maximum degree of attraction as the attraction element. As a result, it can be expressed that the actual customer has finished the product purchase and moved to the accounting machine.
[0087]
Next, in step S23, a movement target is determined. As shown in FIG. 11, since the attracting element 46 is a sales unit and the customer agent 41 cannot enter or leave the attracting element 46, the passive direction of the attracting element (the direction in which the action element can purchase the sales unit that is the attracting element). ) Is set as the movement target 47.
When the attraction element is the accounting unit 15, the movement target is set in the same manner because it is impossible to enter and exit the attraction element. If the attracting element is capable of entering / exiting the action element as in the entrance / exit unit 16, the position where the attracting element is located is set as the movement target.
[0088]
Next, in step S24, it is determined whether or not the obtained maximum attraction level Mmax is a value sufficient to cause the target action. Since the action limit value setting table 7p stores action limit values LM for the action element code, the attraction element code, and the target action code, the determination is made based on these values. If the maximum attraction level Mmax is greater than or equal to the action limit value LM, the process proceeds to the next step S25. Otherwise, the process proceeds to the process of wrinkle behavior in step S28. The action limit value LM is also a value including the estimated value.
[0089]
In step S25, it is determined whether the distance DM between the action element being processed and the movement target is within a range in which the action element causes the target action to occur. Since the action distance limit value setting table 7q stores action distance limit values LDM for action element codes, attracting element codes, and target action codes, determination is made based on these values. If the distance DM between the action element being processed and the movement target is within the action distance limit value LDM, the process proceeds to the next step S26. Otherwise, the process proceeds to the movement action process in step S27. For example, when the purchase behavior is the target behavior, the behavior distance limit value LDM is determined by a reachable range, that is, a so-called reach length. Therefore, although it is an estimated value, a realistic value can be set.
[0090]
In the example shown in FIG. 12, the distance at which the customer agent 41 can purchase the sales unit 46 of the attracting element is zero, that is, the customer agent 41 comes on the movement target 47. Similarly, when the attracting element is the accounting unit 15 or the entrance / exit unit 16, it is determined that the target action can be performed when the customer agent 41 comes on the movement target.
[0091]
In step S26, the target action is performed. Target behavior includes purchasing behavior, accounting behavior, and leaving behavior. In the case of purchasing behavior, the sales unit of the attraction element is the purchased product. Specifically, the number of sales units corresponding to the purchased product is reduced from the status table 7c of the sales unit, and the element code of the sales unit corresponding to the purchased product and the number of purchases are stored in the status table 7e of the customer agent. is there. In the case of accounting behavior, the attracting element is the accounting unit 15 and the position of the customer agent is in the vicinity of the accounting unit, and the processing of the accounting behavior may not end within one time step. In the case of an exit action, when the attracting element is the entrance / exit unit 16 and the position of the customer agent is in the entrance / exit unit, the data table of the exiting customer agent is stored in a storage file of the storage device 6, for example. The customer code corresponding to the processing number of the leaving customer agent is deleted. As a result, the customer agent is not processed until a new store is entered.
[0092]
In step S <b> 27, when the distance DM between the behavioral element being processed (customer agent) and the movement target is larger than the behavioral distance limit value LDM, the behavioral element moves in the direction of the movement target 47. FIG. 12 shows the situation. When moving one cell in one time step, as shown in the trace 48 of the movement action, it moves to the movement target in six time steps and performs a purchase action in the seventh time step. However, since the spatial scanning is performed at each time step during this movement and the visible range changes every time, there is a possibility that the movement target is changed to the movement target of the attracting element having a higher degree of attraction. In addition, since the behavior processing of each behavior element is processed asynchronously with each other, the current attracting element may be purchased as a behavior element to be processed later even within one time step of the current time.
[0093]
In step S28, if the maximum attraction degree Mmax is less than the action limit value LM, the action element (the customer agent does not have a specific movement target and hesitates in the target area 31. At this time, for example, the widest visible range is provided. Rules such as moving in the direction or changing the moving direction at random are set in advance.
This completes the search behavior algorithm processing, and the subsequent processing proceeds to step S14 of the customer agent behavior processing shown in FIG. The exploration behavior described here is a basic behavior pattern, scans the target area (target space), recognizes other target elements that interact with the action element being processed, and It is to evaluate the state in a complex manner and implement the desired action. Thus, this search behavior is used not only by customer agents, but also by employee agents and other examples.
[0094]
FIG. 17 shows an example of an algorithm for accounting behavior. First, in step S31, the accounting mode is determined. That is, when branching by the Mode variable comes to the accounting action for the first time from the search action, the process proceeds to step S32. When the accounting action has already been started, the process proceeds to step S34. In step S32, the total number of products purchased by the customer agent (total number of purchased products) Np is extracted from the customer agent state table 7e. In step S33, accounting is substituted into the Mode variable.
[0095]
Subsequently, in step S34, the number Sr (processing speed) of products that can be accounted per unit time is subtracted from the total number of purchased products Np. That is, the processing speed for the element code of the accounting unit is extracted from the accounting unit state table 7s. This state table 7s is set in relation to the accounting business behavior of the employee agent.
[0096]
Subsequently, in step S35, an accounting action end determination is performed. That is, when there is no unaccounted item number Np, the process proceeds to the next step S36 in order to end the accounting action. However, if the number of unaccounted items still remains, the process proceeds to step S14 of the customer agent action process shown in FIG. Thus, in the next time step, the same customer agent behavior processing in FIG. 9 performs the accounting behavior processing in step S13 from the beginning.
[0097]
In step S36, transaction recording processing is performed. That is, since the accounting action for this customer agent has been completed, the date / time, accounting unit code, accounted customer code, sales unit code, sales quantity (purchased quantity), sales price (purchasing price) are added to the store accounting business status table 7t. Etc. are stored. Therefore, this store accounting business state table 7t corresponds to the sales data in the actual store.
[0098]
In step S37, search mode selection processing is performed, and the Mode variable is changed to search. As a result, in the next time step, in the behavior processing of the same customer agent in FIG. 9, the search behavior processing in step S12 is performed from the beginning.
As described above, since the accounting behavior can be continuously performed through a plurality of time steps, it is possible to faithfully represent the status of accounting behavior that takes a considerable amount of time in an actual store.
[0099]
FIG. 18 shows an algorithm for employee agent behavior processing. The flow of this algorithm is basically the same as that of the customer agent behavior processing algorithm shown in FIG. 9, and the processes in steps S41, S48, and S49 are as follows. The processing is the same as steps S11, S14, and S15 in FIG. Accordingly, here, the six business actions shown in steps S42, S43, S44, S45, S46, and S47 of FIG. 18 will be described.
[0100]
These business behaviors are examples of business operations performed in the store, and may be appropriately increased or decreased. In a relatively small retail store such as a convenience store, for example, two employees may perform all of these six operations. In contrast, the division of roles is clear at relatively large stores such as supermarkets. The business behavior is basically based on the search behavior shown in FIG. 10, and each target business behavior and stationary business behavior are added.
[0101]
In step S42, sales promotion business behavior is performed in the store. This business behavior includes special sales, discounts, installation of POPs, and the like. Among them, the sales promotion business behavior whose date and contents are set in advance by the business plan can be processed in this step, or can be processed in the stationary business in step S8 of FIG. The business performed in step S42 is a sales promotion business that is not determined in advance or whose details are not determined in detail. For example, it is a time service for selling out and discounts a product that is likely to remain unsold in a certain time zone, and is used for products with a short shelf life, such as fresh food and milk.
[0102]
In this sales promotion business behavior processing, an employee agent who performs sales promotion business behavior corresponding to an actual store refers to POS data (sales point-of-sale information) generated from an accounting machine or the like. It is necessary to recognize the remaining status of sales units (products), the congestion status of customer agents, and the purchase status of sales units (products). For this reason, the sales agent (product), the discount amount (discount rate), and the discount time zone where the employee agent moves while recognizing the inside of the store through a search action and recognizes the situation in the store as a whole. Adopts an algorithm to determine.
[0103]
In step S43, an accounting operation is performed in the store. Most of this business behavior is the operation of the accounting machine. There is no need for exploratory behavior when allocating employee agents who constantly perform accounting operations to all accounting units in the store, but in actual stores, the number of accounting machines that operate according to the degree of crowded customers is changed. It is necessary to cope with this. For this reason, in the processing of accounting work behavior, employee agents move while recognizing the inside of the store by scanning behavior and recognize the crowded situation of customers in the store and decide which accounting machine to operate when The algorithm is used. FIG. 6 shows the situation of the employee agent 33a during the accounting operation.
[0104]
In step S44, a product ordering operation action is performed in the store. This business behavior is ordered while confirming the remaining amount of sales units (products) in the store as in the actual store. In other words, the employee agent moves while recognizing the inside of the store by a scanning action through a search action, and recognizes the remaining status of the sales unit (product) in the store and determines the code, quantity, and time of the sales unit (product) to order. The algorithm to decide is adopted. Note that this product order has a set period from when the product is placed to when the product is delivered, so an algorithm that takes this into account may be adopted.
[0105]
In step S45, product replenishment work behavior is performed in the store. In this business behavior, as in an actual store, the remaining amount of sales units (products) in the store is confirmed and a small amount is replenished. That is, the employee agent moves while recognizing the inside of the store by a scanning action by the scanning action, recognizes the remaining state of the sales unit (product) in the store, and replenishes the sales unit (product) with a small remaining amount each time. The algorithm is adopted. FIG. 6 shows the situation of the employee agent 33b during the merchandise replenishment business action.
[0106]
In step S46, a merchandise disposal work action is performed in the store. In this business behavior, the expiration date of the sales unit (product) in the store is confirmed in the same manner as in an actual store, and items whose expiration date expires are discarded. In other words, the employee agent moves while recognizing the inside of the store by a scanning action by the scanning action, grasps the expiration date of the sales unit (product) in the store, and discards the sales unit (product) whose expiration date expires each time. The algorithm that goes is adopted. At this time, contrary to the customer agent action process, the merchandise to be discarded is stored in the store accounting business state table 7t with a negative amount by the accounting machine. That is, it is recorded as loss on the store side.
[0107]
In step S47, a cleaning work action is performed in the store. In addition to the overall cleaning, this business behavior includes a task of confirming a dirty place in the store and cleaning the area with priority, as in an actual store. Here, the employee agent moves while recognizing the inside of the store by spatial scanning by the search action, recognizes the cleanliness of all the target elements in the store, and adopts an algorithm that focuses on the target element and its surroundings is doing.
[0108]
With the above algorithm and data based on the setting of the target area, a program for simulating the behavior of the customer agent and the employee agent in the store can be configured. By executing this program with the simulation computer 1 shown in FIG. In addition, the behavioral simulation of the employee agent in the store can be realized. This simulation model, consisting of this algorithm and the target area, faithfully represents the spatial and temporal conditions in the actual store and the complex interaction between each element as much as possible. Therefore, it is a realistic model, and it is possible not only to reproduce the inside of an existing retail store in the simulation computer 1, but also the spatial and temporal expansion that does not exist in an actual store, between each element. Even if a complex interaction is added, a realistic result can be obtained. For example, it is possible to accurately simulate whether the number of purchased (sold) products increases if the layout in the store is changed, or whether the total price of purchased (sold) products increases if the price of the product is changed. Further, this simulation can be applied to a large store such as a department store having a plurality of floors by connecting the target areas 31 shown in FIG. 6 in a plurality of layers.
[0109]
In this way, the product / service exchange behavior and business behavior generation system can be simulated faithfully to the actual human behavior and the sales state of the product by a relatively simple algorithm.
[0110]
(Second Embodiment)
In this embodiment, a case will be described in which a newly divided space is associated as a subcell with respect to individual cells constituting the target space.
Simulation in the two-dimensional target area in the store is possible in the format shown in FIGS. 6, 11, and 12. However, the actual store space is three-dimensional, and the target area configured there can also be configured in three dimensions by dividing the target space by, for example, cubic (regular hexahedron) cells. The target element is also configured in three dimensions, a more realistic target area is generated, and simulation is basically possible with the same algorithm. In this method, although the fidelity to reality is improved, there are problems such as an increase in calculation time due to a three-dimensional space and an insufficient area of the main memory 5.
[0111]
Therefore, here, the first embodiment is expanded and, for example, a simulation is performed in which sales units can be arranged in the height direction only for the shelf area.
That is, FIG. 19 is a shelf space that is a simulation target when the shelf area is viewed from the front, and is referred to as a target shelf space. Similar to the in-store space, the target shelf space 51 is divided into square cells 52. This divided cell is called a subcell. The position data of each subcell can be set in the same manner as in the store space. However, there is a height direction. In addition to dividing a space in the height direction, the subcell may further divide a flat shelf space arranged in a plane into a plurality of portions.
[0112]
In FIG. 19, the sales unit 53 is already arranged and configured as a target shelf area 54. The target shelf area 54 belongs to each shelf area in the store. The setting of the target element in the shelf space includes a boundary area 55 and a sub shelf area 56 as area elements. The boundary area 55 is a boundary that restricts the movement and visibility of customers and employees, and expresses walls (partition plates), windows (transparent partition plates), and the like. In addition, the sub shelf area 56 represents a space of a fixture in which products, in-store advertisements (POPs) for sales promotion, and the like are arranged. There is a sales unit 53 as a passive element. The sales unit 53 represents products, POPs, and the like, and is arranged on the sub shelf area 56. Furthermore, there are customer agents and employee agents as behavioral elements. A customer agent represents a customer in the store. The employee agent represents the employee in the store.
[0113]
For these target elements, as in the case of the store space, the element code and position data are stored in the shelf space element arrangement table together with the element code of the shelf area in the store to which the target element belongs. The cell size in the shelf space is selected to be smaller than that in the store, and a side of 50 to 150 mm is appropriate. However, if 300 mm is set to be the same as the space in the store, the effect of simplifying the algorithm is great.
[0114]
Next, the shelf searching action in which the sub shelf area is formed in the height direction will be described.
First, considering that the customer agent performs purchase behavior as shown in FIG. 12 in the in-store simulation, in step S26 during the search behavior processing shown in FIG. select. This shelf search behavior is basically the same as the search behavior algorithm in the store shown in FIG. 10 except that the search target space is in the target shelf space. It is what you do within.
[0115]
In the first step, spatial scanning processing is performed. In this spatial scanning processing, the scanning target element is only the sales unit. However, the scanning range differs greatly, and when it was in the store of FIG. 11, it was a substantially fan-shaped scanning range, but here, as shown in the viewable range 58 of FIG. It corresponds to the range of the visual field of the eyes. In FIG. 20, since there is no visual obstacle, the scanning range and the visible range are the same. The space outside the visible range is the invisible range 59. If there is a visual obstacle, a blind spot is generated and the visible range is narrowed.
[0116]
In order to satisfy the size of the scanning range, a plurality of target shelf areas belonging to the shelf area in the store are connected to form a target shelf area to be scanned. The target shelf space shown in FIGS. 19 and 20 is the target shelf belonging to the three shelf areas of the shelf area to which the sales unit 46 of the attracting element in front of the customer agent 41 shown in FIG. Constructed by connecting regions. In addition, in the store search behavior, the center of the scanning range corresponds to the position of the viewpoint center 57 of the behavior element, as the top of the fan-shaped scanning range is the behavior element (customer agent) position in the store.
[0117]
In the next step, the degree of attraction calculation is performed. In this process, the scanning target element is only the sales unit, and therefore the target action is only the purchase action.
In the next step, the movement target is determined, but the attracting element here is visible and can be grasped by hand.
[0118]
Then, similarly to the case shown in FIG. 10, the action limit determination and the action distance limit determination are performed, and any one of the execution of the target action, the movement action, and the habit action is performed. If the target action is to be executed, only the purchasing action is set. Further, if it is a movement action, the viewpoint center of the action element moves toward the movement target. At this time, since the shelf search behavior is only the movement centered on the viewpoint, it can move quite quickly. Also, since the cell is small, one time step in the shelf search action is considerably shorter than one time step in action processing in the store. For example, if one time step in the store is 1 second, one time step in the shelf search action is about 0.1 seconds. In addition, if the behavior is a habit, the habit behavior is almost the same as when the viewpoint center of the action element is in the store.
[0119]
The processing of the shelf search action is completed as described above, but it is not possible to shift to the search action of the in-store simulation being processed as it is. For this reason, for example, when the maximum degree of attraction is lower than a certain limit, the search behavior of the original in-store simulation is shifted to. It is also effective to set a leaving action from the shelf space corresponding to the leaving action.
[0120]
In this way, the in-store simulation to which the shelf search behavior is added can faithfully represent the actual product placement situation. Therefore, it is possible to accurately simulate whether or not the number of purchased (sold) products increases when the product arrangement on the shelf is changed.
[0121]
(Third embodiment)
Both of the above-described two embodiments have described the in-store simulation, but here, the business area simulation will be described.
Stores in an actual retail store belong to a real area called a trade area, and the business and business conditions of the store are affected by a plurality of situations in the trade area. For this reason, in order to comprehensively simulate the management of a retail store, it is necessary to simulate the behavior of customers (purchasers) and employees within the business area to which the target store (target store) belongs.
[0122]
This simulation does not represent the movement of people in the entire business area, the status of products or services, but the status of business and sales at the target store. Since the boundaries of the trade area do not actually exist, the situation of the area outside the trade area has little effect on the business (sales) of the target store, or it is quantified even if the degree of influence does not approach the simulation The basic definition of a trade area is that it can be set manually.
[0123]
Here, the business area simulation will be described.
The business area simulation method is basically the same as the in-store simulation, and includes setting of a target area and a simulation algorithm. The same method as the in-store simulation can be used as a method for setting the target area expressing the business area.
[0124]
A method for setting the target area will be described with reference to FIG. This figure shows a commercial area space that is a target of simulation overlooking the commercial area. This is a target space 61 and represents a two-dimensional space. Similar to the in-store space, the target space 61 is divided by square cells 62. Cell position data and the like can be set in the same manner as in the store space. Here, the target element is already arranged and configured as a target area 63 in the trade area.
[0125]
Next, the setting of the target element in the business area space will be described.
As the area elements, there are a traffic area 64 and a restricted area 65. The traffic area 64 expresses a traffic network such as a road or a railway line on which customers and employees can freely move. Areas other than the traffic area 64, rivers, parks, etc. are expressed in areas where the movement of employees is restricted. Then, one area element is arranged on the square cell 62.
[0126]
As passive elements, there are a resident unit 66, a store unit 67, and an entry / exit unit 68. The resident unit 66 corresponds to a residence where a person in the business area lives, an address to which the person belongs, and a store. The unit 67 represents a place other than a resident unit that exchanges products and services, such as a retail store, a restaurant, a bank, a hospital, a school, an office, a factory, and a station. In addition, the entry / exit unit 68 represents a boundary at which entry / exit between the trade area and the outside is performed. These are usually located in the traffic area. A plurality of passive elements can be arranged in one area element.
[0127]
In addition, although there is a situation of a resident unit and a store unit, here, both the resident unit and the store unit are arranged and expressed in one area element. Spaces such as parking lots basically belong to residential units and store units to which they belong. In addition, passive elements are not arranged in all restricted areas. This part represents a place where goods and services are not exchanged and cannot be freely passed. For example, it represents a restricted area, a vacant house, and a vacant land where the owner is located. It also expresses places that are uninvestigated or impossible to investigate and do not know what is there.
[0128]
The behavioral elements include a customer agent 69 and an employee agent 70. The customer agent 69 is a customer and a customer candidate (one person and one person) in the target store such as a person living in the business area or a person moving to the business area. Expresses the same group). The employee agent 70 represents an employee (one person and a group performing the same action) in the target store. A plurality of passive elements can be arranged in one area element and in the passive element. In addition, when moving using a transportation means such as an automobile or a train, attention is paid only to the movement of a person, and the moving speed is used.
[0129]
Also, Business area simulation Then, since there are cases where a large number of customer agents are handled, customer agents are set by taking advantage of the definition of “group performing the same action”. For example, it is used like a group of students who go home from school. These target elements are stored in the element arrangement state table as element types, element codes, and the like as in the case of the in-store simulation. The position data of the element arrangement according to the actual trade area can be easily constructed from a commercially available house map or an electronic data of it. Customer address settings can also be easily obtained from the customer service card described above.
[0130]
The size of the cell in the target space in the trade area varies depending on the purpose of the simulation, and may be from 1 m to 1 km or other. For example, in order to grasp the visibility in the business area of the target store, about 1 to 3 m is appropriate, and the limit of maintaining the proportional relationship of the area of the store, road, etc. is about 5 to 10 m, the actual area and area Although the proportional relationship disappears, the distance relationship is almost exactly 20 to 60 m.
[0131]
Further, for example, about 100 m to 1 km is used for a large-scale trade area having a radius of 5 km or more using transportation means. In this case, it is necessary to calculate a route from data such as a road map for the distance relationship between elements. Also, for such a large-scale trade area, the linearity and right-angle of buildings and roads are almost irrelevant, so those who use isotropic regular hexagonal cells whose neighboring cells are all in the same condition Is good.
In the case of FIG. 21, since the target store operates mainly for customers who come to the store on foot, a 1620 m square trade area is set with a square cell of about 60 m on a side.
[0132]
The setting of the time step, which is the target time, also varies depending on the size of the cell and the movement speed of the behavioral element. Basically, it is desirable to set so that one cell or more can be moved in one time step. If the cell has a side of 100 m or less, one time step may be set to 1 second to 5 minutes. However, in order to grasp the visibility from the automobile, a setting of 1 second or less may be performed. Even in the case of a cell exceeding 100 m per side, it is within 30 minutes except for special cases. Other settings can be made in the same manner as in-store simulation.
[0133]
Next, the business area simulation algorithm is described. Similar to the in-store simulation, the business area simulation is intended for the simulation of product and service exchange in the target area of the behavioral element and the business generation related thereto. The in-store simulation algorithm can be used with a slight change in settings.
[0134]
First, the overall algorithm of the business area simulation is almost the same as the overall algorithm of the in-store simulation shown in FIG. In the appearance behavior in the trade area, an action element that enters the target trade area from outside the trade area in the entry / exit unit 68 and an action element that starts an action in the target trade area from the residence unit 66 are set. Appearance in the business area is determined with reference to the date and time. For example, school attendance at school, workplace, etc., working hours can be determined in a relatively realistic manner, and shopping time from a house to a store can be set by statistical data or personal data from a customer service card.
[0135]
The behavior processing algorithm of the customer agent in the next business area is almost the same as that of the in-store simulation shown in FIG. 9, and the accounting behavior may be the in-store behavior. Also, the behavior processing algorithm of the customer agent is applied to the behavior processing algorithm of the employee agent in the business area.
[0136]
The algorithm of the search behavior in the business area is substantially the same except for the search behavior of the in-store simulation shown in FIG.
The first change location is a target behavior for the type of attraction element, which is an in-store behavior for the store unit, a return behavior for the resident unit, and an exit behavior for the entry / exit unit. The exit behavior is exactly the same as that in the in-store simulation at the entrance / exit unit. The return home action is exactly the same as the above exit action in the resident unit. In-store behavior, once a customer agent enters a store unit, it stays at that store unit for a predetermined time step and performs a predetermined action. Then, after a predetermined time step elapses, the user exits the store unit. This algorithm is the same as the accounting behavior algorithm in the in-store simulation.
[0137]
The next change is the movement behavior, which is basically the same as in the in-store simulation, but is different in that the behavior element can be selected as appropriate for the transportation means and the corresponding movement speed is set.
With the business area simulation algorithm configured as described above, the behavior of the behavior element in the business area can be expressed faithfully. For example, not only the behavior of going out from a residence and going to a store, but also the behavior of going to a store on the way home from work or school, and the behavior going to a relatively nearby store in rainy weather, the response coefficient to the weather mentioned above is set. It becomes possible by doing. Also, what areas have an effect on competitive relationships between the same business by grasping in which region flyers are distributed will increase the number of customers entering the target store, and by looking at changes in customer store selection? Can be grasped. In addition, if a store is configured at an arbitrary point in the simulation and the number of visitors is expected by the simulation, it will be an effective judgment material when opening a new store.
[0138]
Another purpose of the business area simulation is to combine with the store entry behavior of the in-store simulation.
In the first combination method, first, a business area simulation is performed, and a store entry pattern with respect to the time of the customer agent in the target store is recorded. Next, the customer agent is caused to enter the store based on the store entry pattern recorded in the store entry behavior of the store simulation.
[0139]
The second method is a method of connecting the in-store behavior of the business area simulation with the in-store behavior of the in-store simulation and the exit behavior to form one algorithm as a whole. This method can be the same as the connection of the search behavior of the store space and the search behavior of the shelf space. In other words, the store entry behavior of the in-store simulation may be started with the element code of the customer agent who intends to start the in-store behavior in the business area simulation. Similarly, the in-store behavior may be terminated in the trade area simulation with the element code of the customer agent who intends to perform the exit behavior in the in-store simulation, and the corresponding customer agent may be moved out of the store unit. In this method, since one time step of the simulation is different between the trade area and the store, it is necessary to ensure consistency.
[0140]
As described above, a comprehensive simulation combining the business area and the store becomes possible. Thereby, the fidelity to the reality of simulation can be further improved. In addition, if other stores in the same industry can be obtained with only a simple layout, a plurality of in-store simulations can be combined with the business area simulation to effectively grasp the competitive relationship.
[0141]
(Fourth embodiment)
Next, management of an actual retail store is performed using the simulation of merchandise / service exchange behavior and business behavior generation system related thereto, as exemplified in the in-store simulation and the trade area simulation regarding the retail store in each of the above-described embodiments. A system that supports the decision-making process is explained.
[0142]
Here, the retail store which is the target of management decision-making support is called the target store, and the target store and the environment surrounding it are called the target management system, and it includes the inside of the store and the business area of the actual retail store. Accordingly, simulation of merchandise / service exchange behavior and business behavior generation system is simply referred to as management simulation.
[0143]
FIG. 22 is a block diagram showing the configuration of the management decision support system. The simulation computer 1, the input device 2, and the output device 3 are the same as those shown in FIG.
[0144]
In the figure, reference numeral 71 denotes a target management system, which indicates the inside of a retail store or a commercial area. In the target management system 71, the actual data collecting means collects various actual data. Specifically, the accounting time, customer identification code, product name, product code, product quantity, sales price Point-of-sale data collection device 72 such as a POS system that collects products, product order data collection device 73 such as an electronic ordering system that collects time, product name, product code, product quantity, purchase price, etc. An image data collection device 74 such as a video camera for collecting data relating to the attributes and movements of customers and employees, the arrangement of products, etc. as images is arranged. The image data collection device 74 is connected to an image data processing means 75 comprising, for example, a computer and an image processing program in order to extract data such as attributes such as age and sex of the customer from the image.
[0145]
In addition, weather and temperature data, TV program data, map data in the trade area, traffic data, competing store data, data indicating the management status of the store (target store), Attribute data, customer attribute data, etc. Data required for in-store and business area simulations and data necessary for implementation of the decision support system are collected directly from within the target management system 71, or Collect existing data, such as temperature data and mass-produced product attribute data. Here, all means required for data collection are used as actual data collection means.
[0146]
The simulation computer 1, the point-of-sale data collection device 72, the product order data collection device 73, the image data processing means 75, and the data storage means 76 are connected to a network 77.
[0147]
Real data collected in the target management system 71 is stored in a data storage means 76 such as a magnetic disk via a network 77. The system composed of the simulation computer 1, the input device 2, and the output device 3 makes a management decision making to a user of the management decision support system (hereinafter simply referred to as a user) by means of recommended information generation means that is a computer program. Recommendation information for support is provided. A program that constitutes the recommended information generating means is stored in the storage device 6 and the main memory 5.
[0148]
Next, an example of a functional configuration of the recommended information generation unit will be described with reference to FIG.
The configuration of the function of the computer program of the recommended information generating means is divided into each part of model adapting means 81, management simulation means 82, estimated value setting means 83, various data parts related thereto, and input / output parts.
[0149]
The management simulation means 82 is a simulation program itself for the exchange of goods / services and the business generation system related thereto. The algorithm part of the simulation program is determined by the method disclosed in each of the embodiments described above. The data portion of the simulation program can be obtained from the actual target management system 71, for example, the status value stored in the status table such as the sales unit status table, and the actual target management system 71. Can be classified into estimated values stored in a setting table such as a customer agent to sales unit response coefficient setting table.
[0150]
Furthermore, even if it can be obtained from the actual target management system 71, there is also data such as a behavior element attribute state master table for a customer agent, for example, where it is difficult to obtain all information temporally or economically, These difficult to obtain data are also classified as estimated values.
[0151]
First, the actual state value data formed in the storage device 6 from the actual data directly collected from the point-of-sale data collecting device 72, the product ordering data collecting device 73, and the image data collecting device 74 and the real data extracted from the data storage means 76. Store in table 84. In the actual state value data table 84, data representing the state of the actual management system is stored in time series from the past to the present.
[0152]
The values in the actual state value data table 84 directly correspond to the state values of the management simulation means 82 and are used as the initial state value 85 of the management simulation used when starting the management simulation.
The estimated value setting means 83 generates an initial estimated value 86 of the management simulation used when starting the management simulation from the values in the real state value data table 84 using, for example, a well-known statistical method.
[0153]
The intermediate forced fluctuation state value 87 and the intermediate forced fluctuation value estimated value 88, which are intermediate forced fluctuations in the management simulation, correspond to disturbances to the management system, so whether the user inputs values using the input device 2? Select from preset data. In the execution of the management simulation, the state value and the estimated value are updated from the initial values based on the algorithm every time step, but the intermediate forced fluctuation state value 87 and the intermediate forced fluctuation value estimated value 88 are stored in the algorithm. The state value or the estimated value can be changed to an arbitrary value at an arbitrary time after the initial with a forced fluctuation value that is not based.
[0154]
The final state value 89 and the final estimated value 90 of the management simulation are the final updated values of the state value and the estimated value at the simulation end date and time in the execution of the management simulation, and the situation of the target management system of the simulation end date and time Is expressed.
[0155]
The model adaptation means 81 is the initial stage of the management simulation that is the adaptation target so that the state value 91 as the adaptation target input from the input means matches the final state value 89 of the management simulation that is the comparison target of the adaptation target. All or any combination of the state value 85, the initial estimated value 86, the intermediate forced fluctuation state value 87, and the intermediate forced fluctuation value estimated value 88 are changed. (Adaptation)
By repeating the change of the adaptation target and the execution of the management simulation, a search is performed for a value of the adaptation target that makes it possible to make the matching target and the comparison target coincide with each other or make the values very close. A technique such as a well-known genetic algorithm (GA) is used for the search matching of the value to be adapted. Adaptation is similarly performed on the estimated value 92 as the adaptation target input from the input means. Either one of the adaptation targets or the state value 91 and the estimated value 92 may be entered at the same time to perform the adaptation.
Further, the status of the execution of the simulation by the management simulation unit 82 and the execution of the adaptation by the model adaptation unit 81 are respectively displayed or printed by the output device 3 for the user.
[0156]
Next, an example of a management decision support procedure using the recommended information generating means will be described based on the flowchart of FIG.
Here, the recommended information is generated using real data, and various data related to the management of the target store are derived and output as data used in making a management decision.
[0157]
First, in step S51, the process of adapting the estimated value of the simulation, which is an incomplete value in executing the management simulation, is started by the model adapting means 81.
[0158]
In this adaptation process, first, in step S52, the estimated value is adapted by the model adaptation unit 81 in accordance with the processing conditions. The processing conditions include a simulation period, an initial value state value, an initial value estimated value, an intermediate forced variation, a compliance target, a comparison target, and a calibration target. In addition, intermediate forced fluctuation is not included here.
[0159]
Then, a certain point in the past of the actual management system is set as the starting point of the simulation, and the period from that point to the present is set as the simulation period of the management simulation. Value. The initial estimated value is a preset value. The compliance target is the current state value of a real management system. The comparison target is the final state value of the simulation, and the initial estimated value of the simulation is changed in the adaptation target while comparing with the final state value.
[0160]
In the next step S53, the model adapting means 81 obtains an initial estimated value that can be generated on the simulation based on the current state value from the past state value, thereby obtaining the final estimated value in the simulation from the past to the present. And this is the current estimate of the actual management system. This estimated value is set as the adapted estimated value, and this is used as the initial estimated value in the simulation for the target management system thereafter.
[0161]
Each data collection device 72 to 74 of the target management system 71 shown in FIG. 22 automatically stores data that becomes a state value at a certain time interval, for example, once a day, in the real state value data table 84. In this case, it is possible to set the adaptation processing to be performed automatically, so that the adapted estimated value is automatically updated every day.
[0162]
In the next step S54, condition settings for performing simulation from the current state to the future are set according to the simulation conditions. As simulation conditions, there are a simulation period, an initial value state value, an initial value estimated value, an intermediate forced variation, a conformity target, a comparison target, and an adaptation target. Here, the simulation period of the management simulation is from the present to a certain point in the future input.
[0163]
In the next step S55, the user selects time extrapolation simulation or goal attainment simulation according to the purpose.
When the time extrapolation simulation is selected, the process of time extrapolation simulation is started in step S56. That is, initialization is performed, and changes in the state of the management system based on the passage of time thereafter are expressed on the computer.
[0164]
In step S57, a management simulation is performed according to the condition setting in step S54.
In the condition setting, when the initial state value is the current system state and there is no intermediate forced change, a situation is simulated in which the current target management system continues to elapse as it is. For example, in a retail store, a simulation can be performed in the case where there is no significant change from 1 day to 1 to 3 weeks. Accordingly, in the retail store, for example, it is possible to predict the number of customers entering the store, predict the sales of individual products, etc., and prevent the occurrence of expired products due to out-of-stock or over-ordering, thereby enabling proper product ordering.
[0165]
Also, in the condition setting, if the initial state value is an input value obtained by changing the current system state and there is no intermediate forced change, the situation when a new idea is applied to the current target management system Simulate. For example, in a retail store, when planning sales promotion activities such as special sales days that are not regular, the effect of the planning is the specific management index such as the amount of sales and the number of employees required to sell the product. Can be evaluated, and is effective for verification of ideas.
[0166]
Furthermore, when there is an intermediate forced change in the condition setting, a relatively long-term simulation of one month or longer is targeted. For example, if the production of vegetables and fruits changes significantly in the next few months due to unpredictable effects of warm winter or cold summer, simulation of the target management system when there is a large price fluctuation, before the warm winter or cold summer actually occurs Or it can be done before such a prediction is made, and the situation can be grasped as crisis management.
[0167]
In step S58, the specific situation of the simulation is output to the output device 3, for example, a display, so that the user can confirm the display.
In step S59, the user himself determines whether the displayed situation is satisfactory. If the state value can be satisfied, the process proceeds to step S60. If the state value cannot be satisfied, the process proceeds to step S61. For example, in the condition setting described above, if the initial state is the current system state and there is no intermediate forced fluctuation, and if the sales situation of each individual product at a certain point in the future is satisfactory, the process proceeds to step S60. If not, the process proceeds to step S61. In this step, an initial state for changing the current state of the target management system is set, and the process returns to step S54 again to set the simulation condition, and then the time extrapolation simulation is performed again. .
[0168]
If a satisfactory state is obtained by several changes, the change point for the current target management system state is displayed on the output device 3 in step S60, and the state change is recommended to the user. . At this time, the simulation computer 1 directly transfers data on ordering of products, change of sales price, etc. to these devices via the point-of-sale data collection device 72, the product order data collection device 73 and the network 77, and orders the products. It is also possible to change the selling price.
[0169]
If the goal attainment simulation is selected at step S55, the target attainment simulation process is started at step S62. That is, in the time extrapolation simulation, as shown in steps S59 and S61, the user himself / herself had to set an initial state to change the current state of the target management system. Then, the model adaptation means 81 searches for an initial state candidate that changes the current state of the target management system.
[0170]
First, in step S63, the model adaptation means 81 performs adaptation according to the processing conditions set in step S54. The adaptation target at this time is a value input by the input device 2 or a future state value of the target management system set as a user target set in advance by selection. The comparison target with the conformity target is the final value state value of the management simulation indicating the future state value of the target management system in the simulation.
[0171]
Then, the initial state value of the management simulation indicating the current state value of the target management system in the simulation is changed so that the matching target matches the comparison target. When the total value of sales of individual commodities such as the target amount of total sales is set as the adaptation target, for example, a plurality of initial state values that match the conformance target and the comparison target are derived.
[0172]
In step S64, the model adaptation unit 81 searches for one or a plurality of initial state values that provide the future state value of the target management system, and outputs it to the output device 3, for example, a display. Confirm the display.
In step S65, the user himself determines whether the displayed situation can be satisfied. For example, the initial state value is a change point with respect to the current management system, but if it is difficult to change, the conformity target may be impossible, so the process proceeds to step S67 and the conformance target is changed. Or, if too many initial state values are displayed, the initial condition to be searched is limited, and the process returns to step S54 to set the simulation condition, and then the target arrival simulation is performed.
[0173]
Further, if a satisfactory state is obtained by several changes, the process proceeds to step S66, and at this step, the change point for the current state of the target management system is displayed on the output device 3 to the user. It is recommended to change the state.
This is the end of the contents of an example of the management decision support procedure using the recommended information generating means, but the management decision support procedure using the recommended information generating means is an estimated value which is the basic configuration. It only needs to include the adaptation process, the time extrapolation simulation, and the goal attainment simulation, and is not limited to the procedure of FIG.
[0174]
This management decision support system verifies the user's own ideas regarding specific operations such as store opening plans, management strategies, floor layouts, assortment, shelf allocation, sales promotion, pricing, inventory quantity setting, and product ordering. Can be evaluated using specific management indicators as a scale. In addition, a plurality of specific recommended information can be presented for one management goal, and the range of user selection can be freely changed. It can also be used for strategic issues such as long-term forecasting and crisis management. In this way, it is possible to supply the user with recommended information that supports effective decision making during comprehensive planning and comprehensive evaluation study.
[0175]
Next, a specific example when such a management decision support procedure is applied to an actual retail business will be described.
a. Estimated value adaptation processing
First, estimated value adaptation processing shown in steps S51 to S53 in FIG. 24 is performed.
[0176]
For example, it is assumed that various state values of the target store from one month ago to the present time are stored in the real state value data table 84 of FIG. Here, among the various state values, for example, the area element arrangement state table 35, the passive element arrangement state table 36, the action element arrangement state table 37, the sales unit state table 7c shown in FIG. The customer agent state table 7e shown is taken up.
[0177]
Therefore, the state of the in-store layout at an arbitrary time from one month ago to the present time is expressed by using a part of the area element arrangement state table 35 and the passive element arrangement state table 36, and is shown in FIG. The target area 11 corresponding to the inner layout can be formed. Similarly, the situation of the sales floor structure at an arbitrary time from one month ago to the present time is expressed using the passive element arrangement state table 36, and a target area 21 corresponding to the sales floor structure as shown in FIG. 5 is formed. it can. Furthermore, the state of goods / service exchange behavior and business behavior occurrence in the store at an arbitrary time from one month ago to the present time is expressed using the behavior element arrangement state table 37, and the target area as shown in FIG. 31 is formed.
The sales method status (including the sales promotion status) of each product at an arbitrary time from one month ago to the current time is expressed in the sales unit status table 7c. This is expressed by the customer agent state table 7e.
[0178]
As described above, the state value at an arbitrary time from one month ago to the current time is stored in the actual state value data table 84, and the situation in the store at an arbitrary time from one month ago to the current time can be expressed. However, the estimated values of the customer agent versus sales unit response coefficient setting table 7d shown in FIG. 14 and the customer agent versus purchase behavior response coefficient setting table 7f shown in FIG. 16 are not accurately determined. Therefore, the purpose of the estimated value adaptation process is to accurately adapt the estimated value to the current state value.
[0179]
This adaptation method will be described with reference to FIG. For example, the state value of one month ago is set as the initial state value 85 of the management simulation, and the estimated value setting means 83 is used to set a temporary estimated value one month ago from the state value, and this is used as the initial estimation of the management simulation. The value is 86. As a result, the management simulation means 82 is operated, and a management simulation from one month ago to the present time is performed to calculate an estimated value 92 that is a suitable target. The model adaptation unit 81 compares this value with the state value 91 that is the adaptation target provided by the actual data collection unit. This model adapting means 81 compares the final state value 89 of the management simulation for the current time with the actual current state value, and calculates a temporary estimated value one month ago so that both match with the set accuracy. Revise and run the management simulation again.
[0180]
By repeating this process, it is possible to match the final state value 89 of the management simulation for the current time with the actual current state value with the set accuracy, and to obtain an estimated value one month ago. At the same time, it is possible to obtain a final estimated value 90 of the current management simulation that exactly matches the actual current state value. Thereafter, a management simulation from the present time to an arbitrary future time set as the initial estimated value obtained by adapting the final estimated value 90 of the current management simulation to the current state value is performed. Then, based on a management simulation from the present time to an arbitrary time in the future, some recommended information for current management decision making is supplied to the user as management decision support information.
[0181]
There are two recommended information generation methods, for example, a time extrapolation simulation and a goal attainment simulation, and the conditions and selection are performed in steps S54 and S55 in FIG.
b. Time extrapolation simulation
An example in which the time extrapolation simulation shown in steps S56 to S61 in FIG. 24 is applied to an actual store will be described.
[0182]
In this simulation, the current state value is left as it is, for example, the current value is adopted for the zone element arrangement state table 35 and the passive element arrangement state table 36 to obtain the current sales floor structure, and further sales. The current value is also adopted for the unit state table 7c to obtain the current product sales method. The management simulation means 82 is operated using this and the initial estimated value adapted to the current state value, and the management simulation from the current time to, for example, one week later is performed.
[0183]
As a result of this management simulation, various state values after one week are calculated. For example, the customer agent state table 7e is calculated, the total number of purchased products and the total amount of purchased products for each customer agent are calculated, and this is used as recommended information to be displayed and confirmed to the user via the output device 3. Become.
[0184]
If the user is not satisfied with the recommended information, for example, a change in the price of the product or a change in the discount rate is planned, and the price strength of the sales unit state table 7c The discount rate is changed, and this is set as the initial state value 85 of a new management simulation, and the management simulation from the present time to one week later is performed again.
[0185]
If the user's satisfaction is still not satisfied, the store layout and the sales floor configuration are further planned, and the area element arrangement state table 35 and the passive element arrangement state table 36 are changed to correspond to this, Is set to the initial state value 85 of the new management simulation, and the management simulation from the present time to one week later is performed again.
[0186]
In this way, the above steps are repeated until the user is satisfied, and various state values after one week are finally calculated. The sales unit state table 7c, the zone element arrangement state table 35, the passive element arrangement state table 36, etc. at that time serve as recommended information for the manager of the target store who is the user, that is, management decision support information.
[0187]
c. Objective simulation
An example in which the goal attainment simulation shown in steps S62 to S67 in FIG. 24 is applied to an actual store will be described.
In this simulation, the process of creating a new management plan such as a change in the price of a product in the time extrapolation simulation and the process of repeating the management simulation is performed relatively simply by using the model adaptation means 81 described above. .
[0188]
First, the state value after one week of the target management system, which is the target of the user, is input via the input device 2 as the conformity target.
Next, the current state value is left as it is, for example, the current value is adopted for the area element arrangement state table 35 and the passive element arrangement state table 36, and the current unit value is also applied to the sales unit state table 7c. The value is adopted and used as the current product sales method. The management simulation means 82 is operated using this and the initial estimated value adapted to the current state value, and the management simulation from the present time to one week later is performed.
[0189]
As a result of this management simulation, the customer agent state table 7e after one week is calculated, and the total number of purchased items and the total amount of purchased items are calculated. The model adapting means 81 compares this value with a state value 91 that is a conforming target representing a state value after one week when the user is satisfied. As described above, in the model adaptation unit 81, for example, the area element arrangement state table 35 and the passive element arrangement state table 36 are set so that both (the simulation result and the user setting value) match with the set accuracy. Then, the current state values in the sales unit state table 7c and the like are corrected, and the management simulation is performed again.
[0190]
By repeating this process, it is possible to obtain the current state value that matches the final state value 89 of the management simulation for one week later with the actual state value that the user is satisfied with the set accuracy.
Finally, for example, the current product sales method and the sales floor structure satisfying the total number of purchased products and the total amount of purchased products after one week that the user is satisfied with are calculated, and the corresponding information is stored in the sales unit state table 7c, area It is shown in the element arrangement state table 35, the passive element arrangement state table 36, and the like. This is recommended information for the manager of the target store who is a user, that is, management decision support information.
[0191]
Also, in this goal attainment simulation, setting a state value 91 that is a conforming target that is far from the current state will calculate recommended information that cannot be executed in the actual retail industry, such as an extreme discount rate. Therefore, in this case, it is necessary to change the state value one week after the user who should be satisfied with the state value 91 as the conformity target, and perform this goal simulation again.
[0192]
Next, a case will be described in which the service delivery behavior in the store is simulated using the search behavior algorithm of the in-store simulation shown in FIG. 10 using the product / service delivery behavior and business behavior generation system simulation method.
[0193]
For example, in the case of simulating a product packaging service at a packaging service counter in a retail store, a customer agent's packaging service reception behavior can be simulated using a search behavior algorithm.
In this case, first, the packaging service counter unit is set as a passive element, and is arranged in the target area 11 corresponding to the in-store layout shown in FIG. Therefore, in the passive element arrangement state table 36 of the element arrangement state table 7b, the element code of the packaging service counter unit is set corresponding to the element code of the area element in which the packaging service counter unit is placed.
[0194]
Further, the stimulation value of the packaging service counter unit is set in the target element state table 7c of FIG. Correspondingly, a response coefficient corresponding to the stimulus value of the packaging service counter unit of the customer agent is set in the action element versus target element response coefficient setting table 7d. In addition, a response coefficient relating to the packaging service enjoyment behavior of the customer agent is set in the behavioral element target behavior response coefficient setting table 7f.
[0195]
As for the simulation algorithm of the customer service packaging service reception behavior, the search behavior algorithm of the in-store simulation shown in FIG. 10 can be used. In this case, the degree of attraction of the packaging service counter unit is calculated based on the above setting, and when the attraction element is the packaging service counter unit, the packaging service receiving behavior is set as the corresponding target behavior, As in the case of the first embodiment, the customer agent can move so as to approach the packaging service counter unit and perform the packaging service receiving behavior.
[0196]
Further, the employee agent behavior processing corresponding to this may be performed by adding the employee agent product packaging operation behavior to the employee agent behavior processing algorithm of FIG. Of course, in-store services such as a food service and a food cooking service can be simulated in the same manner.
[0197]
Next, the case of simulating service delivery behavior in the service industry using a product / service delivery behavior and business behavior generation system simulation method will be described.
For example, when simulating a game providing service at a game center, the game providing service enjoyment behavior of a customer agent can be simulated using the in-store simulation algorithm described in the first embodiment.
The simulation method in this case is basically the same as the in-store simulation, and the setting of the target area and the simulation algorithm similar to the above-described in-store simulation can be applied by expressing the game machine as a sales unit.
[0198]
That is, the customer agent searches for the game machine that he / she attracts most in the target area indicating the game center corresponding to FIG. 6, and enjoys the service (game play itself) provided by the game machine. , The behavior of paying the price before or after the service is provided can be simulated. Corresponding employee game machine maintenance work, fee collection work, and the like can be added to the employee agent behavior processing algorithm.
Similarly, other service industries such as a video rental business or a product / service exchange behavior in a store such as a restaurant business can be similarly simulated.
[0199]
Furthermore, regarding the expansion of the application range of the business area simulation, various types of products and services such as retail stores, restaurants, banks, hospitals, schools, offices, factories, and stations, which are shown as store units 67 in FIG. Therefore, it is possible to simulate the selection behavior of a customer agent for a store or a building of such various business types.
[0200]
In each of the above-described embodiments, the management simulation method, that is, the simulation method of the product / service exchange behavior and business behavior generation system, and the management decision support system using this simulation method, the in-store Although the case where the present invention is applied to a business area simulation has been described, the present invention is not necessarily limited to this, and can be widely applied to product / service exchange and business generation systems related thereto. For example, since the situation of the store can be expressed, it can be easily applied to other industries having stores, such as restaurants and service industries. It can also be applied to stores of various sizes, from convenience stores to supermarkets and department stores.
[0201]
It can also be applied to areas where multiple business types such as shopping centers, shopping streets, and underground shopping centers are combined. In that case, the combination of the shopping area and in-store simulation is used, for example, simulation of the whole shopping street and business types. Can easily combine simulations of different stores.
In addition, since it is possible to simulate the relationship between services and users in a specific area, layouts such as amusement centers such as game centers, large leisure centers such as amusement parks, exhibition halls, hospitals, libraries, government offices, stations, etc. Etc.
[0202]
Furthermore, the users of the management decision support system are not only the managers of the target stores, but also wholesalers that distribute products to some stores in a certain region, or manufacturers that manufacture products, Furthermore, financial institutions such as banks can accurately determine the location of the store itself and the composition within the store, and can also be applied to the determination of loans when opening a store or renovating.
[0203]
【The invention's effect】
Book According to the present invention, it is possible to faithfully simulate a product / service exchange behavior and a business behavior generation system with a real person's behavior and a sales state of a product by a relatively simple algorithm. Accordingly, it is effective for grasping the current state of goods and services and business related thereto, improving services, and improving business efficiency. In addition, since the simulation can be configured with the same algorithm from one shelf to the inside of the store, and further to a wide range of trade areas, it is possible to deal with large-scale stores and a plurality of business formats and a combination of these.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a first embodiment of the present invention.
FIG. 2 is a diagram showing a memory configuration of a main memory in the embodiment.
FIG. 3 is a diagram showing an example of cell division of the target space in the embodiment.
FIG. 4 is a diagram showing an example of a target area corresponding to an in-store layout in the same embodiment.
FIG. 5 is a diagram showing an example of a target area corresponding to a sales floor structure in the embodiment.
FIG. 6 is a diagram showing an example of a target area corresponding to the in-store situation in the embodiment.
FIG. 7 is a diagram showing a configuration of an element arrangement state table in the same embodiment.
FIG. 8 is a flowchart showing an algorithm for the entire in-store simulation according to the embodiment;
9 is a flowchart showing an algorithm of customer agent behavior processing in FIG. 8;
10 is a flowchart showing an algorithm for search behavior in FIG. 9;
FIG. 11 is a schematic diagram for explaining spatial scanning and movement target determination in the embodiment.
FIG. 12 is a schematic diagram for explaining the movement behavior and purchase behavior in the embodiment.
FIG. 13 is a diagram showing a configuration of a target element state table used for attracting degree calculation in the embodiment.
FIG. 14 is a diagram showing a configuration of a behavior element versus target element response coefficient setting table used for calculating the degree of attraction in the embodiment;
FIG. 15 is a diagram showing a configuration of a behavior element state table used for attracting degree calculation according to the embodiment;
FIG. 16 is a diagram showing a configuration of a target behavior response coefficient setting table for behavior elements used for calculation of the degree of attraction in the embodiment.
FIG. 17 is a flowchart showing an algorithm of accounting behavior in FIG. 9;
FIG. 18 is a flowchart showing an algorithm for behavior processing of the employee agent in FIG. 8;
FIG. 19 is a diagram for explaining a target shelf space and a target shelf area in the second embodiment of the present invention.
FIG. 20 is a diagram for explaining a scanning range in a target shelf area in the embodiment.
FIG. 21 is a diagram for explaining a target space and a target region corresponding to a trade area space according to the third embodiment of the present invention.
FIG. 22 is a block diagram showing a configuration of a management decision support system in a fourth embodiment of the present invention.
FIG. 23 is a block diagram showing a functional configuration of recommended information generating means in the embodiment;
FIG. 24 is a flowchart showing a management decision support procedure by the recommended information generation means in the embodiment;
[Explanation of symbols]
1. Simulation computer
2 ... Input device
3 ... Output device
4 ... CPU (Central Processing Unit)
5 ... Main memory
6. Storage device

Claims (4)

For the behavior element and the behavior element that performs the product / service exchange behavior and business behavior arranged in the target space consisting of an array of multiple cells, which is the target of the simulation of the product / service delivery behavior and business behavior generation system The passive element that operates passively and the area element that restricts the movement of the behavior element and the arrangement of the passive element, the passive element, the arrangement state table of the area element and the behavior element, and the behavior element of the passive elements state table of the target element comprising a target, said action element of the state table, said pair target element reaction coefficient setting table motion elements, pairs intended behavior reaction coefficient setting table of the motion elements, the scanning capability setting table of the motion elements A memory formed with,
A target area is constructed using data related to the placement of the passive element, area element and action element stored in each arrangement state table in the target space, and the action stored in the scanning capability setting table of the action element The scanning range of the action element in the target region is set from the scanning angle and the scanning distance of the element, and all the passive elements within the scanning range are set as target elements, and the degree of attraction m of the action elements with respect to these target elements is Using the target element state table, the behavior element versus target element response coefficient setting table, the behavior element state table, and the behavior element versus target behavior response coefficient setting table , first, target elements that can be obtained as real data stimulus values are stored in the state table Vi (i = 1,2,3, ···) and, stored in an estimated value pair target element reaction coefficient setting table That reaction coefficient Ci (i = 1,2,3, ···) from the equation [m = Vi · Ci], and then the stimulus value Vsi ( i = 1, 2, 3, ...) Stored in the action element state table available as real data and the estimated value. the reaction coefficients stored pairs purposes behavioral response coefficient setting table Csi (i = 1,2,3, ···) from the equation [m = m- Vsi · Csi] is calculated by suppressing attractant degree m using the element having the greatest attraction of m therein as an attractant element according intended behavior, which is set in advance for the type of the attraction element A simulation computer that changes the data of the state table of the target element and the state table of the action element, and repeats this for a plurality of the action elements;
A management simulation device characterized by comprising: The target space is in a store,
The behavior element is a customer agent;
The target behavior is purchasing behavior, accounting behavior and exit behavior,
When the target behavior preset for the type of the attraction element is the purchase behavior, the simulation computer reduces the number of purchased products from the state table of the passive element corresponding to the attraction element, and The management simulation apparatus according to claim 1, wherein a code and a purchase quantity of the purchased product are stored in a state table of behavior elements corresponding to the agent. The target space is in a store,
The behavior element is a customer agent;
The target behavior is purchasing behavior, accounting behavior and exit behavior,
The memory includes an accounting unit status table and a store accounting business status table;
The simulation computer extracts the total number of products purchased by the customer agent from the state table of the behavior element corresponding to the customer agent when the target behavior preset for the type of the attraction element is the accounting behavior The number of products that can be accounted per unit time stored in the accounting unit state table from the total number of products is repeatedly subtracted until the total number of products becomes zero, and if the total number of products becomes zero, 2. The management simulation apparatus according to claim 1, wherein sales data is stored in the store accounting business state table.   For the behavior element and the behavior element for performing the product / service delivery behavior and business behavior arranged in the target space consisting of an array of a plurality of cells, which is the target of simulation of the product / service delivery behavior and business behavior generation system The passive element that operates passively and the area element that restricts the movement of the action element and the arrangement of the passive element, the arrangement state table of the passive element, the area element, and the action element, and the action element of the passive elements A target element state table, a behavior element state table, a behavior element versus target element response coefficient setting table, and a behavioral element target behavior responsePerson in chargeA memory in which a number setting table, a scanning capability setting table of the behavior elements are formed,
  The target area is constructed using data related to the placement of the passive element, the area element and the action element stored in the respective arrangement state tables in the target space, and the action stored in the scanning capability setting table of the action element The scanning range of the behavioral element in the target area is set from the scanning angle and scanning distance of the element, and all the passive elements within the scanning range are set as target elements, and the degree of attraction of the behavioral elements with respect to these target elementsmUsing the state table for the target element, the target element response coefficient setting table for the action element, the state table for the action element, and the target action response coefficient setting table for the action element.First, the stimulus value Vi (stored in the target element state table available as real data is used. i = 1,2,3, )) And the response coefficient Ci () stored in the target element response coefficient setting table which is an estimated value. i = 1,2,3, )) From the equation [m = Vi · Ci], and then the stimulus value Vsi (stored in the action element state table available as real data) i = 1,2,3, )) And the response coefficient Csi (the target action response coefficient setting table stored as an estimated value). i = 1,2,3, )) From the formula [m = m- Vsi · Csi] to suppress the degree of attraction mCalculatedTheThe greatest degree of attraction among themmAnd changing the data of the state table of the target element and the state table of the action element according to a target action set in advance for the type of the attraction element. A simulation means for repeatedly performing an element;
  Real data collection means for collecting various state values necessary for simulation from actual product / service exchange behavior and business behavior generation system,
  A real state value data table for storing various state values collected by the real data collecting means;
  At the start of simulation by the simulation means, the various state values stored in the real state value data table are the arrangement state tables of the passive element, area element and action element, the state table of the target element, and the state table of the action element. Means for securing the initial state value of
  Means for ensuring a final state value in the state table of the behavior element at the end of the simulation by the simulation means;
  Means for displaying the final state value;
Management decision support system characterized by having

Images