JP2010152504A - Commodity shelf arrangement device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To generate a commodity arrangement state when arranging, in a shelf, commodities displayed in a store in consideration of an evaluation value corresponding to sales profit. <P>SOLUTION: The commodity shelf arrangement device includes: a convenience information setting part generating convenience information of a commodity corresponding to obtained profit in each commodity and each shelf position when each of the plurality of commodities having different kinds is arranged any of the shelf positions; a genetic arrangement generation part executing prescribed calculation to a plurality of commodity arrangement states belonging to a first initial individual group and an N-th (N≥1) individual group, and repeatedly performing processing of generating an N+1th individual group including a plurality of new commodity arrangement states; and an arrangement determination part calculating the evaluation value wherein the convenience information of the respective commodities arranged in all the shelf positions are added to the respective commodity arrangement states belonging to the first and N+1th individual groups, and determining the optimum commodity arrangement state based on the evaluation value. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a product shelf placement device, and more particularly to a product shelf placement device that can determine product placement in consideration of conditions such as sales profit when many products are placed on a plurality of shelves.

  The way products are displayed at convenience stores and other stores has a significant impact on sales. In other words, it is known that sales differ depending on which product is placed at which position in the store and at which height the product is placed in a shelf having a plurality of shelf positions. In order to improve sales, various arrangements have been made for product arrangement.

For example, in order to make it easy for customers to find products, lead wires for guiding customers from the store entrance to the store and the layout of shelves are devised.
Further, in the shelves for displaying products, a device has been devised such that products that sell well are placed at a shelf position near the center so that they are easily noticed, and items for children are placed at a lower shelf position so that they are easy to pick up.

Furthermore, a support system that provides information for a person in charge to determine which product should be placed on which shelf, and various methods and systems for so-called shelf allocation have been proposed.
For example, in Patent Document 1, a variety of information relating to product shelf allocation is created in consideration of product attributes and sales data, and the arrangement of products on the shelf is displayed so that the user can consider the sales allocation data. A method for creating a shelf allocation that provides information for studying is described.
Further, in Patent Document 2, a barcode of an already placed product is read, the barcode is received by a plurality of receivers, and a display shelf in a store is stored by a computer using information such as product data and a display shelf. A system for creating product layout information has been proposed.
JP 2002-230127 A JP 2003-16161 A

In the conventional system as described above, a person in charge of a store can obtain information that is useful for examining shelf allocation of products in the store. However, specific shelf allocation based on sales, that is, where the product is placed to improve sales profit, depends on various factors such as store circumstances, season, region, and sales time zone. Based on the above, the reality is that it is determined from the intuition and experience of the person in charge.
Although it is possible to improve sales to some extent by shelving products based on the experience of such personnel rather than placing the products without any consideration, is shelving optimal based on this experience? Whether or not is unknown. In addition, from the viewpoint of sales profit, there may be an appropriate arrangement of products other than the shelf allocation. Furthermore, there is no one that provides a reasonable way to determine whether the shelf allocation is appropriate using numerical data such as sales performance.

  Therefore, the present invention has been made in consideration of the above-described circumstances, and by placing a product on a shelf in a store, a sales profit can be maximized by using a formulated algorithm. It is an object of the present invention to provide a product shelf arrangement device that generates a product arrangement state that can be performed.

  The present invention is a product shelf placement device that places a plurality of different types of products on a shelf having a plurality of shelf positions, and sets information on the products to be placed and the number of shelf positions where the products are to be placed. If the specific product is arranged at any one of the shelf positions, the benefit information of the product corresponding to the obtained profit is stored in the past for each product and each shelf position. A benefit information setting unit that is generated from sales performance information, and a plurality of product placement states in which an arbitrary product among the plurality of products is temporarily placed at the shelf position are generated, and the generated plurality of product placement states are the first An initial arrangement setting unit for setting the initial individual group, and a predetermined operation for the product arrangement state belonging to the first initial individual group and the Nth (N ≧ 1) individual group, N + 1th individual with new product placement A genetic arrangement generation unit that repeatedly performs a process of generating a group, and benefit information for each of the products arranged in all the shelf positions with respect to each of the generated product arrangement states belonging to the first and N + 1th individual populations The product shelf arrangement device is provided that includes an arrangement determination unit that calculates an evaluation value obtained by adding and determines an optimum product arrangement state based on the evaluation value.

  According to this, the benefit information generated based on the product placement position is used to calculate the evaluation value corresponding to the sales profit, and the optimum product placement state is determined using this evaluation value. From the viewpoint of sales profit, it is possible to generate a product arrangement state capable of improving the sales profit as compared with the product arrangement state determined based on the experience and intuition of the person in charge as in the past.

  In addition, the genetic arrangement generation unit is arranged at every shelf position from among the benefit information generated by the benefit information setting unit for each of the product arrangement states belonging to the Nth individual population. Benefit value information of each product is extracted, an evaluation value calculation unit that calculates an evaluation value obtained by adding the benefit information of all the extracted placement products, and any two of the product placement states belonging to the Nth individual group Select two product placement states and generate two new product placement states by replacing the products placed in one or more corresponding shelf positions with each other for the two selected product placement states Changing a product arranged at an arbitrary shelf position to another product with respect to the crossover processing unit and a predetermined number of product arrangement states selected from the product arrangement states belonging to the Nth individual group A mutation processing unit for generating a new product placement state, a product placement state belonging to the Nth individual group, and a new product placement state generated by the crossover processing unit and the mutation processing unit. A selection processing unit that selects a plurality of product arrangement states based on the evaluation value and generates an N + 1-th individual group that includes a new combination of the selected product arrangement states. .

According to this, the crossover process, the mutation process, and the selection process are executed on the product arrangement state belonging to the individual population of a certain generation (Nth) using the evaluation value in consideration of the sales profit, and the next generation Since the (N + 1) th individual group is generated, it is possible to generate an excellent product arrangement state from the viewpoint of sales profit.
Here, the crossover process, the mutation process, and the selection process may be executed using a data processing method known as a so-called genetic algorithm (GA).

Moreover, in this invention, you may make it the said selection process part select a predetermined number of goods arrangement | positioning state in an order from the thing with the said high evaluation value among several goods arrangement | positioning states used as selection object.
Further, the benefit report of the product is multiplied by the sales probability indicating the possibility of sale when the product is placed at a specific shelf position and the profit at the time of sale of the product obtained when the product is sold. The numerical values obtained may be used.

Also, as one embodiment of the present invention, the arrangement information setting unit initially sets the number of repetitions of processing executed by the genetic arrangement generation unit, and the genetic arrangement generation unit sets the set number of repetitions. Only the predetermined arithmetic processing is repeatedly executed, and the product placement state having the largest evaluation value among the product placement states belonging to the last individual group generated after the placement determination unit has executed the number of repetitions, The optimum product arrangement state may be determined.
Furthermore, as another embodiment, the placement information setting unit initially sets a placement determination condition to be compared with the evaluation value, and the placement determination unit determines that the evaluation value of the product placement state belonging to the generated individual population is When the arrangement determination condition is satisfied, the product arrangement state may be determined as an optimum product arrangement state.

Further, in the present invention, the initial arrangement setting unit may include a plurality of items on the condition that only one item of the same type is arranged at one shelf position and the item of the same type is arranged only at one shelf position. You may make it arrange | position goods to each shelf position.
Alternatively, the initial placement setting unit places a plurality of products on each shelf position on condition that only one product of the same type is placed in one shelf position and the same type of product is placed in a plurality of shelf positions. You may make it arrange | position to.

  Further, the present invention is a product shelf placement program for placing a plurality of different types of products on a shelf having a plurality of shelf positions, the computer trying to place the product information and the products to be placed If the arrangement information setting unit for setting the number of shelf positions to be arranged and the specific product is arranged at any of the shelf positions, the benefit information of the product corresponding to the obtained profit is set for each product and A plurality of benefit information setting units that are generated from past sales performance information for each shelf position, and a plurality of product placement states in which any product among the plurality of products is temporarily placed at the shelf position An initial placement setting unit that sets the product placement state of the first initial individual population, and a predetermined operation for the product placement state belonging to the first initial population and the Nth (N ≧ 1) population Run multiple new A genetic layout generation unit that repeatedly performs a process of generating an (N + 1) th individual population having a product layout state, and all shelf positions for each product layout state belonging to the generated first and (N + 1) th individual population An evaluation value obtained by adding the benefit information of each of the products arranged in the product is calculated, and based on this evaluation value, a product shelf arrangement program is provided for functioning as an arrangement determination unit that determines an optimal product arrangement state. is there.

  According to this invention, the benefit information corresponding to the profit when the product is placed on the shelf position is generated for each product, and a predetermined calculation is performed on the plurality of product placement states belonging to the individual group. For each new product placement state generated, an evaluation value is calculated using the benefit information, and the optimal product placement state is determined based on this evaluation value. From the viewpoint of sales profit, it is possible to generate a more preferable product arrangement state than the product arrangement state determined based on the intuition.

Hereinafter, the present invention will be described in detail based on the embodiments shown in the drawings. In addition, this invention is not limited by this.
<Configuration of Product Shelf Arrangement Device of the Invention>
This invention makes a computer perform the process which arrange | positions several goods to several shelf position which comprises the shelf installed in the shop. Therefore, the product shelf arrangement device of the present invention is realized by a computer such as a personal computer or a workstation. The computer mainly includes a microcomputer including a CPU, a ROM, a RAM, an I / O interface, a timer, a storage device such as a hard disk, a storage medium such as a CD-ROM, a display device such as an LCD and a CRT, and a keyboard. And input devices such as a mouse.

  Also, each functional block of the above-described product shelf arrangement device of the present invention, that is, an arrangement information setting unit, a benefit information setting unit, an initial arrangement setting unit, a genetic arrangement generation unit, an arrangement determination unit, an evaluation value calculation unit, a crossover process The functions of the unit, the mutation processing unit, and the selection processing unit are realized when the CPU organically operates each hardware based on the control program of the apparatus stored in the ROM or the like.

  The control program is for causing the computer to function as each of the above functional blocks, and may be provided in a state stored in advance in a ROM or the like, but is provided by being stored in a storage medium such as a CD-ROM or USB memory. May be. Furthermore, the control program itself may be stored in an external server or the like, downloaded to a computer via a communication line such as the Internet or a LAN, and stored in the hard disk or the like.

FIG. 10 shows a block diagram of a configuration of an embodiment of the commodity shelf arrangement device of the present invention.
Referring to FIG. 10, the product shelf arrangement apparatus of the present invention also includes a display unit 1 such as an LCD, an input unit 2 including a keyboard and a mouse, and a storage unit 3 such as a hard disk, as with a computer such as a general personal computer. Prepare.
In addition, the present invention is characterized by comprising an initial setting unit 4, a genetic arrangement generation unit 5, and an arrangement determination unit 6.
Furthermore, the control part 10 which consists of CPU for controlling operation | movement of each said functional block is provided.

In FIG. 10, the initial setting unit 4 is a part for setting information necessary for arranging the products on the shelf, and mainly includes an arrangement information setting unit 41, a benefit information setting unit 42, and an initial arrangement setting unit 43. Consists of.
The genetic layout generation unit 5 mainly includes an evaluation value calculation unit 51, a crossover processing unit 52, a mutation processing unit 53, and a selection processing unit 54.

The arrangement information setting unit 41 is a part for setting at least information on the products to be arranged and the number of shelf positions existing on the shelf on which the products are to be arranged.
Here, the product information is a product sold at a certain store and is information on a target product to be placed on a shelf. For example, information indicating the type of the product, a product code, and the number of target products (described later) Variable N).
The number of shelves may be one or more, but the number of shelves (a variable m described later) is also set. The number of shelf positions (a variable L, which will be described later) constituting one shelf means the number of stages of placement locations on one shelf.
The process of the arrangement information setting unit 41 corresponds to a coding process to be described later.

The benefit information setting unit 42 is a part that generates benefit information for each product and each shelf position from past sales performance information.
Here, the benefit information (hereinafter, also simply referred to as “benefits”) refers to profits that can be obtained if one particular product among the products to be placed is placed at one of the shelf positions. The corresponding information is information generated for each shelf position arranged for one product.
For example, when the product a is arranged in any of the five shelf positions (1-5), the benefit information P _{a1 at} the shelf position ₁ , the benefit information P _{a2 at} the shelf position ₂ , and the benefit information P at the shelf position 3 _A total of five pieces of benefit information are generated for the product a for each shelf position (1-5), such as _a3 .

An example of this benefit information is shown in FIG.
The benefit information is generated from past sales performance information. For example, the benefit information is obtained when a certain product is placed at a specific shelf position and indicates the probability of sale, and when that product is sold. A numerical value calculated by multiplying the profit at the time of sale of the product (sales price-cost) may be used.
In addition to the multiplication as described above, the benefit information may be calculated by defining another calculation formula, or may be directly input as a numerical value obtained by the user of this device based on his / her own standard. Good.
The process of the benefit information setting unit 42 corresponds to a part of an initialization process described later.

The initial arrangement setting unit 43 is a part that generates an initial individual group (also referred to as a first initial individual group) that is a first individual group, and corresponds to an initialization process that will be described later.
Here, the initial individual population means a population including a plurality of product arrangement states (a variable M described later), and corresponds to a chromosome set described later.
In addition, an initial product group including a plurality (M) of product placement states in which any product selected from a plurality of products to be placed is temporarily placed at any of a plurality of shelf positions. It is. One product arrangement state in which a plurality of products are temporarily arranged at all shelf positions corresponds to one chromosome X described later.
Therefore, it is preferable that the plurality of product arrangement states constituting the initial individual group are different from each other in the arrangement location of any of the products or the products arranged in the shelf positions.
Such initial generation of the plurality of product arrangement states is preferably performed at random, but the person in charge may intentionally generate them.

The genetic arrangement generation unit 5 performs a predetermined operation on the product arrangement state belonging to the first initial individual group generated as described above, and generates an individual group consisting of a plurality of new product arrangement states. It is a part that generates something different from the initial population.
Further, the above-described predetermined calculation is repeated to generate an N + 1-th individual group consisting of a plurality of new product placement states from the product placement state belonging to the Nth (N ≧ 1 integer) individual population It is. This individual group number N means the number of generations (t) described later.
The predetermined operation corresponds to data processing determined by a so-called genetic algorithm (hereinafter referred to as GA), and specifically, crossover processing, mutation processing, and selection processing as described later. , Means a series of processing consisting of evaluation value calculation processing.

The genetic arrangement generation unit 5 includes four functional blocks (51 to 54) that execute each of the series of processes.
Here, the evaluation value calculation part 51 is a part which calculates the evaluation value which added the benefit information of all the goods (arranged goods) arrange | positioned at the shelf position. This evaluation value corresponds to a target function value J described later. For example, the evaluation value is calculated for each of the product arrangement states belonging to the Nth individual group.

In addition, for each of the product placement states, the benefit information generated for each product placed at all the shelf positions is extracted from the benefit information generated by the benefit information setting unit 42, and for each placed product An evaluation value is calculated by adding all the benefit information extracted in the above.
For example, if a certain product a is placed at shelf position 3 for one product placement state, the benefit information generated by the benefit information setting unit 42 (the benefit table in FIG. 6 described later) The benefit information corresponding to the profit obtained when the product a is arranged at the shelf position 3 is extracted.

Similarly, benefit information specified by the product and the shelf position is also extracted for the placed product placed at another shelf position.
For example, if the number of shelf positions is 20, a total of 20 pieces of benefit information are extracted, and the sum of the extracted 20 pieces of benefit information is an evaluation value of the product arrangement state. .
If there are 100 product arrangement states belonging to the Nth individual group, there are also 100 evaluation values calculated for this Nth individual group.

  The crossover processing unit 52 is a part that performs processing corresponding to the crossover defined by the so-called GA. For example, any two product placement states (A and B) are selected from the product placement states belonging to the Nth individual group, and one or a plurality of corresponding product placement states are selected. This is a part that generates two new product placement states (A ′ and B ′) by replacing the products placed at the shelf positions with each other.

Here, when the product arrangement state A selected first and the newly generated product arrangement state A ′ are compared, only the products arranged at one or a plurality of specific shelf positions are different. A product that is a different product and is placed at the shelf position in the product placement state A ′ is a product that was originally placed in the selected product placement state B.
The two product placement states (A, B) selected in this way correspond to parent chromosomes described later, and the product placement state (A ′, B ′) newly generated after replacement is stored in the child chromosomes described later. Equivalent to.

The mutation processing unit 53 is a part that performs processing corresponding to a mutation defined by a so-called GA.
For example, a predetermined number (one or more) of product placement states (C) are selected from the product placement states belonging to the Nth individual group, and the products placed at any shelf position for the state C are selected. This is a part for generating a new product arrangement state (C ′) by changing to another product selected from the products to be arranged.
Specifically, if a certain product d is arranged at the shelf position 2 in one product arrangement state, this means that the product d is changed to another product e.
Here, the shelf position for changing the product may be arbitrary and may be selected at random. However, the selection may be made based on a predetermined rule.

The selection processing unit 54 is a part that performs processing corresponding to selection determined by so-called GA.
Here, a product arrangement state with a number smaller than M is selected from M product arrangement states belonging to a certain individual group. The selection criterion may be random, but it is preferable to select a product arrangement state that is superior in terms of sales profit. Therefore, the above evaluation value is used as a criterion for selection.
For example, based on the evaluation value calculated for each of the product placement states from among the product placement states belonging to the Nth individual group and the new product placement states generated by the crossover processing unit and the mutation processing unit, Select multiple product placement states.
The number selected here corresponds to a variable P described later. A new combination of the selected plurality of product arrangement states is generated as the (N + 1) th individual group corresponding to the next generation.

The usage of the evaluation value serving as a criterion for selection is not limited to a specific one.
For example, an evaluation value may be calculated for each of M product arrangement states belonging to the Nth individual group, and a predetermined number (variable P) of product arrangement states may be selected in order from the highest evaluation value. This selection method corresponds to a method called an elite strategy in GA. Alternatively, a method called roulette selection in GA may be used.

The arrangement determining unit 6 of the present invention shown in FIG. 10 determines a product arrangement state that is considered to be optimal from the individual population generated after repeatedly performing the arithmetic processing determined by the genetic arrangement generating unit 5. Part. The evaluation value described above can also be used for this determination.
For example, for each of the product placement states belonging to the generated first and N + 1 individual populations, an evaluation value obtained by adding the benefit information of each product placed at all the shelf positions is calculated. Is determined as the optimal product placement state.

In order to obtain a product arrangement state that is more favorable in terms of sales profit, this determination process is preferably performed after the arithmetic process by the genetic arrangement generation unit 5 is repeated a predetermined number of times.
The predetermined number of repetitions is not uniquely determined, and may be set to a numerical value that is empirically predicted that the evaluation value will converge to a specific value, for example. This number of repetitions corresponds to the maximum generation number gmax described later.

Alternatively, the arrangement information setting unit 41 initially sets an arrangement determination condition that can be compared with the evaluation value, and each evaluation value of the product arrangement state belonging to the generated individual group satisfies the arrangement determination condition. The product arrangement state may be determined as the optimum arrangement state.
For example, if a product arrangement state having an evaluation value larger than the threshold value appears, the product arrangement state can be obtained by a simple method as the arrangement determination condition. What is necessary is just to determine as an arrangement | positioning state.

Further, in the present invention, in the following embodiments, mainly on the condition that only the same type of product is arranged in one shelf position, and the same type of product is arranged only in one shelf position, A plurality of products shall be arranged at each shelf position. For example, if one or more products a are arranged at one shelf position 3, then another product b cannot be placed at shelf position 3, and product a is placed at another shelf position 2. Cannot be placed.
However, as will be described later, each product may be arranged on condition that the same type of product is arranged at a plurality of shelf positions.

<Product placement in stores>
A lot of display fixtures are arranged in the store. A single display fixture (hereinafter also referred to as a shelf) is configured with a plurality of places (hereinafter referred to as shelf positions) where products are placed.
FIG. 1 shows an embodiment of a product shelf configuration model.
Here, one shelf is configured from five shelf positions. There are various heights of shelves, but generally those having a height of 135 cm, 150 cm, and 180 cm are often used.

Generally, a shelf is considered to have the following three functions.
(1) Merchandise display function (2) Inventory function (3) Display function The merchandise display function means that the customer can easily take out the merchandise, and for example, the range from 45 cm to 150 cm from the floor is the display effective range. Has been. In particular, as shown in FIG. 1, the shelf position (3, 4) in the range from 75 cm to 135 cm is the position where the product is most easily seen and taken out.
Further, the space at a position higher than 150 cm (shelf position 5) is a position that performs the above-described display function, and the product itself is rarely arranged. Furthermore, a position lower than 45 cm (shelf position 1) is often used as a position for placing a heavy product or fulfilling the above-mentioned product inventory function.

In the present invention, when a plurality of products are arranged on a shelf having a plurality of shelf positions as shown in FIG. 1, an optimization method known as a genetic algorithm (GA) is used.
In the following embodiments, in order to facilitate the description, the case where there is one shelf and the case where there are two or more (the number of shelves ≧ 2) will be described separately.

<Description of Genetic Algorithm (GA)>
A genetic algorithm is a method in which humans and other living species artificially create a natural fit and natural selection method, and repeatedly apply it to the target event (for example, product shelf arrangement). By doing this, it is a method of trying to find the optimal solution (for example, the product shelf arrangement with the best sales profit).

  Species generally inherit genetic information every time they change generations, but they are not exactly the same but are changed and inherited. At this time, genetic information changes due to the mechanism of survival of the fittest and mutation. The genetic algorithm GA is formulated by mimicking such changes in the genetic information of species.

The GA mainly includes (1) coding processing of a problem to be processed, (2) initial population generation processing (initialization), (3) crossover processing, (4) mutation processing, (5) selection processing, ( 6) It is comprised from each process of an evaluation process (evaluation value calculation).
Here, a process of crossover, mutation, and selection is applied to an individual population of a certain generation (for example, t generation), and an individual population of the next generation (t + 1 generation) is generated. Then, in the evaluation process, a predetermined evaluation value (a target function value described later) is calculated for this next generation (t + 1), and it is evaluated whether or not this next generation satisfies a target criterion. If not satisfied, the crossover, mutation, and selection processes are repeated for the next generation (t + 1) again, and the individual population of the next generation (t + 2) is generated.

By repeating such processing, new generations are generated one after another, and if the standard is satisfied in a certain generation, the processing is terminated assuming that the generation is the most appropriate solution. Alternatively, when it is determined that the calculated evaluation value converges and does not improve any more, the process is terminated and the generation at that time is determined as the optimum solution.
When such GA is applied to the product shelf arrangement, it is necessary to define and encode the elements necessary for the product shelf arrangement as variables for the sake of formulation.
In addition, an individual population of a generation with GA corresponds to one having a plurality of one product arrangement state in which all target products are arranged at predetermined shelf positions. Furthermore, one chromosome referred to as GA corresponds to one commodity arrangement state, and one gene of GA corresponds to one shelf position arrangement state.
Hereinafter, the Example which applied GA to merchandise shelf arrangement | positioning is shown.

<Encoding process>
(1) When there is one shelf FIG. 2 shows a numerical representation of the shelf positions that constitute one shelf.
In FIG. 2, this shelf has L shelf positions, and the shelf position variable is k. That is, k = 1, 2,..., L.
Further, the shelf position at the bottom of the shelf is k = 1, and the shelf position k at the uppermost stage is L.

Further, the variable of the product to be placed at the shelf position k is i, and the number of types of the product to be placed is N. Here, i = 1, 2... N.
In this embodiment, it is assumed that only the same type of product i is arranged in one shelf position k.

と定義する。 Benefits of placing product i at shelf position k

It is defined as

The benefits are
Benefit = Probability of sales x Profit at the time of sale (1)
Shall be pointed to.
Whether the product i is placed on the shelf position k is x _{i, k} , x _{i, k} = 1 when placed, and x _{i, k} = 0 when not placed.
In addition, the following two conditions are adopted as conditions for the product placement.
Only one product can be placed in one shelf position.
In addition, the product shall not be placed in more than one shelf position depending on whether it is placed on any shelf position.

このようにコード化された商品棚配置の各変数や便益について、次式で表される目標関数値Ｊ（x_p(t)）を計算する。 When the above definitions and conditions are coded (formulated), it is expressed by the following formula.

A target function value J (x _{p (t)} ) expressed by the following equation is calculated for each variable and benefit of the product shelf arrangement thus coded.

次々と生成される各世代の個体集団ごとに前記したようなＧＡの処理を適用した後に、この目標関数値Ｊを計算し、評価する。
一般に、目標関数値Ｊは、ある値に収束するように変化することが多いので、たとえば、ＧＡの各処理を繰り返し行った後、目標関数値Ｊをその都度求め、Ｊの値がこれ以上大きくならないようになった場合の世代の個体集団を、最適な商品棚配置として決定する。

After the GA processing as described above is applied to each individual population of each generation generated one after another, this target function value J is calculated and evaluated.
In general, the target function value J often changes so as to converge to a certain value. For example, after each GA processing is repeated, the target function value J is obtained each time, and the value of J is larger than this. The individual population of the generation when it becomes impossible to be determined is determined as the optimal product shelf arrangement.

とする。
商品ｉを棚ｊの棚位置ｋに置くか置かないかをx_i,k,jとし、置くときをx_i,k,j＝1、置かないときをx_i,k,j＝0とする。
また、上記した棚１つの場合と同様に、棚位置には商品は1つのみ置けるものとする。商品はどれかの棚位置に置くか置かないかで、2ヶ所以上の棚位置には置かないものとする。 (2) If there are m shelves,
Here again, it is assumed that there are N types of products i to be arranged (i = 1,..., N).
Also, as shown in FIG. 3, it is assumed that there are a total of m shelves j (j = 1, 2,..., M). Each shelf j has Lj (j = 1 = 1,..., M) shelf positions k (k = 1,..., Lj).
Also, the benefits of placing product i on shelf position k (k = 1, ..., Lj) of shelf j (j = 1, ..., m)

And
X _{i, k, j} is set whether or not the product i is placed on the shelf position k of the shelf j, x _{i, k, j} = 1 when placed, and x _{i, k, j} = 0 when not placed .
Further, as in the case of one shelf described above, only one product can be placed at the shelf position. The product shall not be placed in more than one shelf position, depending on whether it is placed on any shelf position.

As described above, when there are m shelves, the above definitions and conditions are formulated and expressed by the following equation.

そして、棚が１つの場合と同様に、ＧＡの各処理を繰り返し、Ｊが最大値となったときの世代の商品配置状態を、最適な商品配置状態とすればよい。
以上のような商品棚配置の各変数の定義や条件の定式化を行って、ＧＡの各処理を実行する。 In this case, a target function value J expressed by the following equation is calculated.

Then, as in the case of one shelf, each process of GA is repeated, and the product placement state of the generation when J reaches the maximum value may be set to the optimum product placement state.
Each of the GA processes is executed by defining the variables of the product shelf arrangement and formulating the conditions as described above.

In this embodiment of the product shelf arrangement, the specific values of each variable are defined as follows.
Number of target products: N = 20 (i = 1, ..., 20).
Number of shelf positions: L = 5 (k = 1, ..., 5).
Number of shelves: m = 2 (j = 1, 2)
Variable x _{i, k, j} = 1 or 0.
At this time, the number of commodity variables placed at all the shelf positions is 200 (= 20 × 5 × 2), and there are 200 variables x _{i, k, j} .

Here, the population (hereinafter referred to as chromosome) X indicating the variables x _{i, k, j} is defined as follows.

For example, if x _1,1,1 = 1, this means that the product with product number i = 1 is placed on the shelf position with shelf position number k = 1 in the shelf with shelf number j = 1. .
If x _20,2,1 = 0, it means that the product (i) 20 is not placed in the shelf position (k) 2 in the shelf with the shelf number (j) 1.
Therefore, one chromosome X is composed of numerical values of 0 and 1, for example (1,0,0, ... 1, ... 0), and corresponds to one commodity arrangement state. Further, one variable x _{i, k, j} constituting a chromosome corresponds to one gene.

<Initialization process>
Here, an initial individual population having M chromosomes X as described above is generated. For example, M = 100 may be set and randomly generated. However, it produces | generates so that all the conditions of above-mentioned Numerical formula (6), (7), (8) may be satisfy | filled.
Each of the chromosomes belonging to the initial population (M chromosomes) generated in this way corresponds to the first generation, and indicates one product placement state in which the target product is provisionally placed at the appropriate shelf position. is there. Since the possibility that the initial value of such an arrangement is an optimum value is very low, a product arrangement state that is considered to be optimum is generated by repeatedly applying GA.

Also, in the initialization process, one embodiment of a benefit table composed of the above-described benefit values is generated.
For example, FIG. 6 shows an embodiment of the benefit table when the product (lot i) is placed at each shelf position (k = 1 to 5) of each shelf (j = 1, 2). Each numerical value in the table indicates a benefit value. Here, for example, if a product with i = 1 is placed at shelf position (k) 1 of shelf (j) 1, the benefit of that product (i = 1) is 26.
Further, when the same product (i = 1) is arranged at shelf position (k) 3 of shelf (j) 2, the benefit of the product (i = 1) is 80.
In FIG. 6, among products i, i = 1 to 10 are women's products, i = 11 to 17 are men's products, and i = 18, 19, and 20 are children's products.

The numerical value of this table is set from the sales performance information of the product as described above. However, as far as this table is seen, the product of i = 1 has the greatest benefit when placed at shelf position 3 of shelf 2. In other words, in this case, the sales profit of the product (i = 1) is the largest.
In addition, such a benefit table is based on past product arrangements at the store and sales performance information such as sales quantity and price, and the store staff will modify the store as necessary in consideration of the store circumstances. You can also make it.

<Evaluation value calculation process>
Here, the target function value J defined by the above equations (5) and (9) is calculated.
This target function value J corresponds to an evaluation value used for evaluating the product shelf arrangement. The target function value J is calculated for each of the product arrangement states belonging to the Nth individual group, for example.
Also, from the benefit information generated by the benefit information setting unit (benefit table in FIG. 6), the benefit information of each product placed on all the shelf positions is extracted, and the benefit information of all the placed products that have been extracted. The evaluation value obtained by adding is the target function value J.
This evaluation value calculation processing is executed for each product arrangement state belonging to the first initial individual group. The process is also executed for each product arrangement state belonging to the Nth individual group. At this time, it is also executed for a new product arrangement state generated by the crossover process and the mutation process.

<Selection process>
Here, several chromosomes with a high target function value are selected from among a plurality of chromosomes (product arrangement state) that are currently evaluated.
For example, the target function value J of each chromosome is calculated for the initial population of 100 chromosomes described above, and 20 (= P) chromosomes are selected in order from the largest among the function values J. .
Here, the number P to be selected is not particularly limited to P = 20, and may be an arbitrary value.
In the case of P = 20, about 20% of M = 100 chromosomes having a high evaluation value are selected. However, the chromosomes having the same structure, that is, the merchandise arrangement state having the same merchandise shelf arrangement should not be selected redundantly.

<Chromosome crossover>
Here, in a living body, the same processing is performed as when two chromosomes cross to generate two chromosomes of the next generation progeny. In general, there is a processing method such as one-point crossover with only one point selected at random for variable x in chromosome X, or multipoint crossover with two or more points.
Specifically, from the chromosomes selected by the selection process, any two chromosomes (product arrangement state) are selected as parent chromosomes, and the corresponding variable part in the two chromosomes is replaced (this) Are called crossover) to generate two new child chromosomes (product arrangement state).

  Speaking of the product shelf arrangement, the product arrangement state (A, B) of two different shelf arrangements is selected as the parent arrangement, and the products at the corresponding shelf positions in the two parent arrangements (A, B) are selected. This corresponds to generating two child arrangements (A ′, B ′) by replacing each other. In this crossover process, in general, when one-point crossover is used, the crossover rate Pc may be set to 0.7.

<Gene mutation treatment>
Here, for a chromosome selected by the selection process, an arbitrarily selected gene among the genes (variable x) included in the chromosome is changed.
For example, if an arbitrary variable x _{i, j, k} composing the chromosome X is selected and the numerical value is originally 1, the value is changed to 0.
In other words, the product shelf arrangement corresponds to changing the shelf position where a product is placed to another position.

This mutation process is a process necessary to prevent convergence to a non-optimal local solution due to the GA iteration process. For example, if the GA process is repeated without performing the mutation process at all, the probability that a new chromosome will be generated by crossover will be reduced, and the generated chromosome will be biased. The chance of being generated is much lower.
That is, the mutation process is an effective method for correcting such a bias.
The position and number of genes causing mutations are not particularly limited, and may be selected randomly, for example. The mutation rate Pm indicating the rate of causing mutation may be, for example, Pm = 0.01.

<Example of GA processing procedure>
FIG. 4 shows an embodiment of the processing procedure of the genetic algorithm used in the product shelf arrangement of the present invention. Here, as described above, one chromosome corresponds to one product arrangement state of the product shelf arrangement.
Step 1 corresponds to the encoding process described above. Here, a numerical value is set for each GA variable corresponding to the product shelf arrangement.
Step 2 corresponds to the initialization process described above. Here, the initial solution matrix means the above-described initial individual population, and M so-called first generation chromosomes X are generated.
In step 3, the above-described target function value J is calculated for the initial solution matrix, that is, the first generation chromosome belonging to the initial individual population.
In the first generation, the function value J is calculated for the product arrangement states (chromosomes) belonging to all the initial solution matrices generated in step 2.

The function value J calculated here is an evaluation value for each chromosome, and it can be said that generally a high evaluation value is preferable as a target chromosome.
In the product shelf arrangement, this target function value J is calculated using the benefits for each product extracted from the benefit table, and each product is placed at the shelf position indicated by its chromosome (product arrangement state). It corresponds to the overall evaluation value of and corresponds to the sales profit.

  Therefore, obtaining as large a function value J as possible is equivalent to increasing sales profit. However, since it is difficult to obtain the maximum function value J by performing each process of GA only once, in the present invention, a series of GAs is further added to the obtained next-generation chromosome (product arrangement state). In order to obtain a more appropriate chromosome (product arrangement state) from the viewpoint of sales profit.

In step 4, it is determined whether or not the number of generations t currently generated is larger than the initially set maximum generation number gmax.
In principle, the most appropriate solution is when the function value J is maximum. However, in general, it is known that this function value J converges to a constant value (maximum value) by repeating GA over a plurality of generations, so that it is appropriate if GA is repeated an appropriate number of times. It is thought that a solution can be obtained.

Therefore, in step 4 of the flowchart of this embodiment, the processing is performed as many times as possible (the maximum generation number gmax) at which such an appropriate solution can be obtained.
As the maximum generation number gmax, an arbitrary number can be set. For example, when the convergence tendency as shown in FIG. 8 is shown, gmax = 50 may be set.

In step 5, the crossover process as described above is performed, and two new child chromosomes are generated from the selected two parent chromosomes.
In step 6, the mutation process as described above is performed to generate a chromosome in which a gene at a specific position (arrangement state at a certain shelf position) is changed.
In step 7, the target function value J described above is calculated.
Here, the targets for calculating J are the chromosomes generated in steps 5 and 6 and other chromosomes belonging to the individual population of that generation.
In step 8, the above-described selection process is performed, and a predetermined number (P) of chromosomes (product arrangement state) are selected from the currently targeted chromosomes. Chromosomes that have not been selected (that is, the product arrangement state) are deleted. As a selection criterion, for example, an “elite strategy method” for selecting a predetermined number (P) of chromosomes in descending order of the function value J may be employed.
After step 8, the process returns to step 4, and when the process is not terminated, the processes of steps 5 to 8 are repeated for the selected new chromosome group (individual population).
Note that the order of step 5 and step 6 may be reversed.

<Evaluation of product shelf arrangement>
Here, an example of a result of applying GA to a case where products having the benefit table as shown in FIG. 6 are arranged at a total of ten shelf positions is shown.
When 20 items are arranged on two shelves (1, 2) each having five shelf positions (1-5), information indicating one item arrangement state is as shown in the above equation (10). Is represented by chromosome X having 200 variables.
However, since the notation method of equation (10) is complicated, a variable yi as shown in FIG. 5 is defined, and a notation method in which the variables x _{i, j, k} are converted into the variable yi is used.
FIG. 5 shows an embodiment of this variable conversion rule.

Here, the converted chromosome is represented by Y consisting of ₂₀ variables yi (y _{1 to} y ₂₀ ).
Y = (y ₁ , y ₂ , ..., y ₂₀ ) (13)
In FIG. 5, y ₁ is i = 1, the variable x _i corresponding to k = 1 and j = _1, which means the variables that aggregate to _1,1. Thereby, 200 variables x can be expressed by 20 fewer variables yi.

FIG. 7 shows an example of the initial solution matrix (initial individual population) generated in Step 2.
This is a specific example of a part of the chromosome X represented by the formula (10), which is indicated by Y.
Corresponding to ₁₀₀ chromosomes (x ₁ ,..., X ₁₀₀ ), a part of 100 converted chromosomes (Y ₁ , Y ₂ ... Y ₁₀₀ ) is shown.
The above-described GA processing is applied to 100 such chromosomes Y.

FIG. 8 shows a graph of the change in the target function value J calculated for this example.
The vertical axis in FIG. 8 is the numerical value of the target function value J, and the horizontal axis indicates the generation number of the generated chromosome. In addition, among the two line graphs in the figure, the upper line graph indicates the numerical value of the product placement state having the maximum target function value J among a plurality of product placement states belonging to an individual population of a generation. Yes. The lower line graph shows the average value of the target function values J of a plurality of product arrangement states belonging to an individual population of a certain generation.
In FIG. 8, in the first generation, the value of J was about the maximum value of 650. However, by repeatedly applying GA, the value of J gradually increases, and in the 43rd generation, the value of J is the maximum. It is about 710, and it can be seen that it has almost converged.
That is, even if the maximum generation number gmax is set to a numerical value of 50 or more, it can be seen that only the same maximum value can be obtained as the value of J.

このＹは、J＝712であり、第４３世代の商品配置状態のうち、最も評価値の高い商品配置状態を示している。 In the case of FIG. 8, the target function value J is maximized when the number of generations is 43, and the chromosome Y at this time is expressed as follows.

This Y is J = 712, and indicates the product placement state with the highest evaluation value among the 43rd generation product placement states.

Here, there are five positions (k = 1 to 5) as the shelf position k, and the shelf position 1 is the lowest level. Of the two shelves, the shelf 1 is on the entrance side of the store and the shelf 2 is on the back side of the store.
In this state, the first value “−3” of Y of the optimal solution means that the first product (i = 1) is arranged at the position of k = 3 and j = 2. That is, it means that it is preferable to place the product (i = 1) at the shelf position 3 on the shelf (j = 2) on the back side of the store.
The 10th value “1” of Y indicates that the 10th product (i = 10) is k = 1 and j = 1, that is, shelf 1 on the entrance side of the store, and the shelf position below that 1 means that it is preferably arranged.

Further, the eleventh numerical value “−4” of Y indicates that the product (i = 11) is arranged at the position of the shelf position 4 of the shelf with the shelf number 2.
Hereinafter, similarly, the twelfth numerical value “−5” of Y indicates that the product (i = 12) is arranged at the position of shelf position 5 of the shelf with shelf number 2.
The 15th numerical value “5” of Y indicates that the product (i = 15) is placed at the position of shelf position 5 of the shelf with shelf number 1.
The 16th numerical value “3” of Y indicates that the product (i = 16) is arranged at the position of shelf position 3 of the shelf with shelf number 1.
The 17th numerical value “4” of Y indicates that the product (i = 17) is placed at the position of shelf position 4 of the shelf with shelf number 1.
The 18th numerical value “−2” of Y is for placing the product (i = 18) at the position of shelf position 2 of the shelf with shelf number 2;
The 19th numerical value “−1” of Y is for placing the product (i = 19) at the position of shelf position 1 of the shelf with shelf number 2.
The 20th numerical value “2” of Y means that the product (i = 20) is arranged at the position of shelf position 2 of the shelf with shelf number 1.

Further, in the optimal product arrangement state of this embodiment, since there are only 10 shelf positions, the second to ninth products (i = 2 to 9) are not arranged on the two prepared shelves. means.
FIG. 9 shows an explanatory diagram of one embodiment of the product arrangement state of the present invention.
The numbers in the figure are the lot i numbers shown in FIG. 6. For example, 1 to 10 indicate products for women.
FIG. 9A is an example of a product arrangement state in which the evaluation value (target function value) J is 688, and is an arrangement that appears around the 19th generation to the 23rd generation.
FIG. 9B shows an example of a product arrangement state in which the evaluation value J is 706, which is the arrangement of the highest points generated in the 38th and 39th generations.
FIG. 9C shows a product arrangement state of the 43rd generation highest evaluation value (J = 712). According to this, it is understood that as the number of generations increases, an arrangement with a high evaluation value is gradually generated.
Further, regarding the products for women, four are arranged in FIG. 9A, but three products are shown in FIG. 9B and two products are shown in FIG. 9C. Therefore, in this embodiment, from the viewpoint of sales profit, it can be said that it is better to arrange a large number of men's products than women's products.

FIG. 11 shows an example of product shelf arrangement in the case where products are arranged on shelves according to the experience and intuition of the person in charge as in the prior art. Here, a case where 20 items are arranged on two shelves (1, 2) having five shelf positions is shown.
11 (a) and 11 (b), the person in charge selects 10 products out of 20 products by his / her own intention, and selects the selected product from the product characteristics, sales performance, etc. Is arranged in consideration of Let shelf 1 be a shelf close to the entrance and shelf 2 be a shelf in the back of the store. Further, it is assumed that there are products for children (18 to 20), women (1 to 10), and men (11 to 17).

In FIG. 11 (a), considering such product types, children's products (18, 20) are placed at the lowest shelf position 1, and women are placed at the upper shelf position (2, 3). It shows the product (1,4,5,10) arranged for.
Further, from the experience of the person in charge, inexpensive products are arranged on the shelf 1 near the entrance, and more expensive products than the products placed on the shelf 1 are arranged on the shelf 2 in the back of the store.

  In the case of FIG. 11A, the target function value J calculated by the equation (9) was 644. This numerical value is lower than the maximum value (712) of the final function value J obtained by applying GA to the same case.

Further, in FIG. 11 (b), children's products (18, 19) are arranged at the lowest shelf position 1 of the two shelves (1, 2), and the shelf position (2, 3) of the shelf 1 is displayed. For ladies (7,1), shelves (4,5) for men (16,14), shelves 2 for other shelves (2-5) The case where the product for women is arranged is shown.
This FIG. 11 (b) is also arranged based on the experience of the person in charge, but here it is considered that men generally tend to spend less time selecting products than women. In order to facilitate selection of products of gender, they are arranged together on one shelf.

In the case of FIG. 11B, the target function value J calculated by the equation (9) was 658. This numerical value is also lower than the maximum value of the function value obtained by applying GA.
Therefore, it can be seen that when the product arrangement is generated by applying the GA, the evaluation value J considering the sales profit can be generated higher than the product arrangement performed by the person in charge based on experience and intuition.
That is, by repeatedly performing a series of processes to which GA is applied, it is possible to generate an arrangement state of products that can improve sales profit more than before.

In the above embodiment, the same type of merchandise is applied only to one shelf position, and GA is applied under the condition that it is not placed in two or more different positions.
However, in an actual store, the same product is often arranged at a plurality of different shelf positions.
For example, products that sell well, seasonal products, etc. are often arranged near the entrance or conspicuous apart from the positions that are usually arranged.
In such a case, it is necessary to apply GA under the conditions that allow overlapping arrangement in order to place one product or the same type of product at a plurality of shelf positions. In the case where the overlapping arrangement is allowed, it is only necessary not to adopt the expressions (4) and (8) among the conditional expressions described above. Thereby, it is possible to obtain an appropriate solution for the product shelf arrangement when the same type of products are arranged in an overlapping manner.

It is explanatory drawing of one Example of the structural model of the goods shelf of this invention. In this invention, it is explanatory drawing of the shelf position in case there is one shelf. In this invention, it is explanatory drawing of the shelf position in case there are a plurality of (m) shelves. It is one Example of the process sequence of the genetic algorithm (GA) of this invention. It is explanatory drawing of one Example of the variable conversion rule of this invention. It is explanatory drawing of one Example of the benefit table of this invention. It is explanatory drawing of one Example of the initial stage individual population of this invention. It is the graph of one Example which showed the change of the target function value of this invention. It is explanatory drawing of one Example of the goods arrangement | positioning state of this invention. It is a block diagram of the configuration of an embodiment of the commodity shelf arrangement device of the present invention. It is explanatory drawing of one Example of the merchandise shelf arrangement | positioning which the person in charge arranged conventionally.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Display part 2 Input part 3 Memory | storage part 4 Initial setting part 5 Genetic arrangement | positioning production | generation part 6 Placement determination part 41 Placement information setting part 42 Benefit information setting part 43 Initial arrangement | positioning setting part 51 Evaluation value calculation part 52 Crossover process part 53 Mutation Processing unit 54 Selection processing unit k Shelf position j Shelf number

Claims (9)

A product shelf placement device that places a plurality of different types of products on shelves having a plurality of shelf positions, and information on the products to be placed and the placement information setting for setting the number of shelf positions on which the products are to be placed And
If the specific product is arranged at any one of the shelf positions, the benefit information of the product corresponding to the obtained profit is generated from the past sales performance information for each product and each shelf position. An information setting section;
An initial placement setting unit that generates a plurality of product placement states in which any of the plurality of products is temporarily placed at the shelf position, and sets the generated plurality of product placement states in a first initial individual group; ,
A predetermined operation is performed on the product arrangement state belonging to the first initial individual group and the Nth (N ≧ 1) individual group, and an N + 1th individual group composed of a plurality of new product arrangement states is obtained. A genetic arrangement generation unit that repeatedly performs the generation process;
An evaluation value obtained by adding the benefit information of each of the products arranged in all the shelf positions is calculated for each of the product arrangement states belonging to the generated first and N + 1 individual groups, and based on this evaluation value A product shelf arranging device comprising: an arrangement determining unit for determining an optimal product arrangement state. The genetic arrangement generation unit includes:
For each of the product placement states belonging to the Nth individual group, the benefit information of each product placed in all the shelf positions is extracted from the benefit information generated by the benefit information setting unit, and extracted An evaluation value calculation unit that calculates an evaluation value obtained by adding the benefit information of all placed placement products,
Arbitrary two product arrangement states are selected from the product arrangement states belonging to the Nth individual group, and the two selected product arrangement states are arranged at one or a plurality of corresponding shelf positions. A cross-processing unit that generates two new product placement states by replacing the products with each other;
For a predetermined number of product placement states selected from the product placement states belonging to the Nth individual group, a new product placement state can be obtained by changing a product placed at an arbitrary shelf position to another product. A mutation processing unit to be generated;
Based on the evaluation value, a plurality of product placement states are selected from the product placement state belonging to the Nth individual group and the new product placement state generated by the crossover processing unit and the mutation processing unit. The product shelf arrangement device according to claim 1, further comprising: a selection processing unit that selects and generates an (N + 1) th individual group including a new combination of the selected product arrangement states.   The product shelf placement device according to claim 2, wherein the selection processing unit selects a predetermined number of product placement states in descending order of the evaluation value from among a plurality of product placement states to be selected. .   The benefit information of the product is a numerical value obtained by multiplying the sales probability indicating the possibility of sale when the product is arranged at a specific shelf position by the profit at the time of sale of the product obtained when the product is sold. The product shelf arrangement device according to any one of claims 1, 2, and 3, wherein   The arrangement information setting unit initially sets the number of repetitions of the process executed by the genetic arrangement generation unit, and the genetic arrangement generation unit repeatedly executes a predetermined calculation process for the set number of repetitions, The placement determining unit determines the product placement state having the largest evaluation value as the optimum product placement state among the product placement states belonging to the last individual group generated after executing the number of repetitions. The product shelf arrangement device according to claim 1 or 2.   When the placement information setting unit initially sets a placement determination condition to be compared with the evaluation value, and when the placement determination unit satisfies the placement determination condition, the evaluation value of the product placement state belonging to the generated individual population The product shelf arrangement device according to claim 1, wherein the product arrangement state is determined to be an optimal product arrangement state.   The initial arrangement setting unit places a plurality of products at each shelf position on the condition that only one product of the same type is arranged at one shelf position and the same type of product is arranged only at one shelf position. The product shelf arrangement device according to claim 1, wherein the product shelf arrangement device is arranged.   The initial arrangement setting unit arranges a plurality of products at each shelf position on condition that only one product of the same type is arranged at one shelf position and the same type of product is arranged at a plurality of shelf positions. The product shelf arrangement device according to any one of claims 1 to 6, wherein: A product shelf placement program for placing a plurality of different types of products on a shelf having a plurality of shelf positions, the computer comprising:
Information on the products to be arranged, an arrangement information setting unit for setting the number of shelf positions where the products are to be arranged,
If the specific product is arranged at any one of the shelf positions, the benefit information of the product corresponding to the obtained profit is generated from the past sales performance information for each product and each shelf position. An information setting section;
An initial placement setting unit that generates a plurality of product placement states in which any of the plurality of products is temporarily placed at the shelf position, and sets the generated plurality of product placement states in a first initial individual group; ,
A predetermined operation is performed on the product arrangement state belonging to the first initial individual group and the Nth (N ≧ 1) individual group, and an N + 1th individual group composed of a plurality of new product arrangement states is obtained. A genetic arrangement generation unit that repeatedly performs the generation process;
An evaluation value obtained by adding the benefit information of each of the products arranged in all the shelf positions is calculated for each of the product arrangement states belonging to the generated first and N + 1 individual groups, and based on this evaluation value A product shelf arrangement program for functioning as an arrangement determination unit for determining an optimal product arrangement state.

Images