JP7754730B2 - Demand curve estimation device and ordering support system - Google Patents

Description

This disclosure relates to a demand curve estimation device and an ordering support system.

Dynamic pricing has been attracting attention in recent years. Dynamic pricing refers to determining a sales price that maximizes gross profit based on the sales price of a commodity, such as a product or service, and an estimated demand for that commodity. When dynamic pricing is implemented in a situation where sufficient sales volume of a commodity at each of multiple different sales prices has not been accumulated, it is necessary to update prices while balancing "exploration" and "exploitation." "Exploration" refers to setting a price at which the gross profit is unclear and accumulating sales data. "Exploration" refers to setting a price that maximizes gross profit. Various technologies have been proposed for updating prices while balancing "exploration" and "exploitation." For example, Patent Document 1 discloses a technology that estimates a demand curve using a contextual bandit algorithm.

Japanese Patent Application Laid-Open No. 2021-163047

The contextual bandit algorithm is an algorithm that has been used in the advertising field, for example, to estimate the most effective web articles based on the characteristics of the web article's viewers when placing web advertisements. When the contextual bandit algorithm is applied to estimating demand curves, it has been found to produce inappropriate demand curves. Specific examples of inappropriate demand curves include demand curves that show overall demand that is too high or too low for each of multiple selling prices, or demand curves that are not downward sloping. If the demand curve is inappropriate, gross profit cannot be maximized.

This disclosure has been made in light of the above circumstances, and aims to solve the problem of maximizing gross profit in situations where there is insufficient sales volume of merchandise at each of multiple different sales prices.

A demand curve estimation device according to one aspect of the present disclosure includes an estimation unit, an acquisition unit, and a correction unit. The estimation unit estimates a demand curve indicating the expected demand for a commodity, which is a good or service, relative to its selling price, based on sales performance indicating the sales volume of the commodity at each of a plurality of different prices. The acquisition unit acquires a first demand volume calculated from the demand curve for a first selling price and a second demand volume calculated from the demand curve for a second selling price different from the first selling price. When the second selling price is lower than the first selling price and the first demand volume is greater than the second demand volume, the correction unit corrects the demand curve to raise the second demand volume to or above the first demand volume; and when the first selling price is lower than the second selling price and the second demand volume is greater than the first demand volume, the correction unit corrects the demand curve to lower the second demand volume to or below the first demand volume.

An ordering support system according to one aspect of the present disclosure includes the above-described demand curve estimation device and an order quantity determination device. The order quantity determination device acquires the inventory quantity of the merchandise in the store. The order quantity determination device determines the order quantity of the merchandise based on the acquired inventory quantity and the demand curve corrected using the demand curve estimation device. The ordering support system of this aspect maximizes gross profit in situations where the sales volume of the merchandise at each of multiple different sales prices is not sufficiently accumulated.

This disclosure makes it possible to maximize gross profit even in situations where there is insufficient sales volume for merchandise at each of multiple different sales prices.

1 is a diagram illustrating a configuration example of an order support system 1 according to an embodiment of the present disclosure. FIG. 1 is a diagram illustrating an example of the configuration of a demand curve estimation device 10. FIG. 10 is a diagram showing an example of a demand curve GY estimated by the estimation unit 111. FIG. 10 is a diagram showing an example of a demand curve GYA1 estimated by the estimation unit 111. FIG. 10 is a diagram showing an example of a demand curve GYA2 estimated by the estimation unit 111. A diagram showing the relationship between the demand curve GY and the upper limit of the confidence interval for each selling price. 10A and 10B are diagrams for explaining a first correction process executed by a correction unit 114. FIG. 10A and 10B are diagrams for explaining a second correction process executed by a correction unit 114. FIG. 10A and 10B are diagrams for explaining a third correction process executed by a correction unit 114. FIG. 10 is a flowchart showing the flow of a demand curve estimation method executed by the processing device 110 according to the program P. 1 is a flowchart showing the flow of an order quantity determination method executed in the order support system 1.

<A. Embodiment>
Fig. 1 is a diagram illustrating an example configuration of an order support system 1 according to an embodiment of the present disclosure. The order support system 1 is an information system for supporting ordering of merchandise sold at stores 40A and 40B. As shown in Fig. 1, the order support system 1 includes a demand curve estimation device 10 and an order quantity determination device 20. As shown in Fig. 1, each of the demand curve estimation device 10 and the order quantity determination device 20 is connected to a communication network 30, such as the Internet.

In this embodiment, each of stores 40A and 40B is a retail store such as a convenience store. Store 40A sells product X, such as sweets. Store 40B sells product X and product Y, such as soft drinks. As shown in FIG. 1, terminal device 41A is installed in store 40A. Terminal device 41B is installed in store 40B. Terminal device 41A and terminal device 41B are each connected to communication network 30. Terminal device 41A and terminal device 41B each have a communication function for communicating with order quantity determination device 20 via communication network 30.

At store 40B, a sales clerk (hereinafter referred to as clerk B) inputs information indicating the type and quantity of merchandise sold into terminal device 41B. In response to the input by clerk B, terminal device 41B transmits sales performance information indicating the type and quantity of merchandise sold, as well as a second identifier indicating store 40B, to the order quantity determination device 20. The sales quantity is the number of merchandise sold. Similarly, at store 40A, a sales clerk at store 40A (hereinafter referred to as clerk A) inputs information indicating the type and quantity of merchandise sold into terminal device 41B in response to the sale of merchandise. Terminal device 41A transmits sales performance information and a second identifier indicating store 40A to the order quantity determination device 20 in response to the input by clerk A.

The order quantity determination device 20 stores the order quantity of each product type (in other words, the amount purchased at each store) in association with each of stores 40A and 40B. The order quantity determination device 20 also stores sales performance information received from each of stores 40A and 40B via the communication network 30. In this embodiment, the demand curve for each product is estimated by the demand curve estimation device 10 for each store based on the sales performance information stored in the order quantity determination device 20.

A demand curve is a curve that shows the number of commodities (demand quantity) that are expected to be sold for a given price. In this embodiment, the demand curve is estimated by the demand curve estimation device 10 at a predetermined time after the closing of business for the day at stores 40A and 40B, for example. The order quantity determination device 20 determines the purchase quantity and sales price that maximize gross profit based on the demand curve estimated by the demand curve estimation device 10 and the inventory quantity of the commodities. Gross profit is the value obtained by subtracting sales costs such as purchase prices from the sales amount. Inventory quantity is the value obtained by subtracting the sales quantity indicated by the day's sales performance from the previous purchase quantity.

As mentioned above, product Y is not sold at store 40A, so when store 40A starts selling product Y, the demand curve is estimated using the sales history of product Y at store 40B. However, if the period between the start of sales of product Y at store 40B and the start of sales of product Y at store 40A is short, there may not be sufficient sales history for product Y. In this embodiment, by having the demand curve estimation device 10 execute a demand curve estimation method that prominently demonstrates the features of the present disclosure, an appropriate demand curve can be obtained even when sufficient sales history has not been accumulated.

Figure 2 is a diagram showing an example configuration of the demand curve estimation device 10. As shown in Figure 2, the demand curve estimation device 10 includes a processing device 110, a communication device 120, and a storage device 130. The processing device 110 is, for example, a CPU (Central Processing Unit). The processing device 110 functions as the control center of the demand curve estimation device 10 by operating in accordance with a program stored in the non-volatile storage device 132. The communication device 120 is a device that communicates with the order quantity determination device 20 via the communication network 30.

As shown in FIG. 2, the storage device 130 includes a volatile storage device 131 and a non-volatile storage device 132. The volatile storage device 131 is, for example, RAM (Random Access Memory). The volatile storage device 131 is used by the processing device 110 as a work area for executing programs. The non-volatile storage device 132 is, for example, flash ROM (Read Only Memory). Various programs are stored in the non-volatile storage device 132. Specific examples of programs stored in the non-volatile storage device 132 include a kernel program and program P. The kernel program is a program that causes the processing device 110 to implement an OS (Operating System). The kernel program is not shown in FIG. 2. Program P is a program that causes the processing device 110 to implement a demand curve estimation method that prominently demonstrates the features of the present disclosure.

When the power supply (not shown in FIG. 2) of the demand curve estimation device 10 is turned on, the processing device 110 reads the kernel program from the non-volatile storage device 132 to the volatile storage device 131. The processing device 110 starts executing the kernel program read into the volatile storage device 131. The processing device 110, operating in accordance with the kernel program, implements the OS. The processing device 110, while implementing the OS, reads the program P from the non-volatile storage device 132 to the volatile storage device 131 when the aforementioned predetermined time is reached. The processing device 110 then starts executing the program P read into the volatile storage device 131. The processing device 110, operating in accordance with the program P, functions as an estimation unit 111, a determination unit 112, an acquisition unit 113, and a correction unit 114, as shown in FIG. 2. The functions of the estimation unit 111, determination unit 112, acquisition unit 113, and correction unit 114 are as follows:

The estimation unit 111 estimates a demand curve for each store and product based on the sales performance of the product. More specifically, the estimation unit 111 estimates a demand curve for product X in store 40A based on the sales performance of product X in store 40A. The estimation unit 111 estimates a demand curve for product X in store 40B based on the sales performance of product X in store 40B. The estimation unit 111 estimates a demand curve for product Y in store 40B based on the sales performance of product Y in store 40B. Furthermore, the estimation unit 111 estimates a demand curve for product Y in store 40A based on the sales performance of product Y in store 40B.

If the estimation unit 111 is able to obtain sales records for product Y at each of multiple stores, it may estimate a demand curve for product Y at store 40A based on the sales records for each of the multiple stores. Furthermore, the estimation unit 111 may estimate a demand curve for product Y at store 40A based on the sales records for product Y at a store among the multiple stores that is the same as or similar to store 40A in at least one of store size, location, and customer demographic. This estimation improves the accuracy of the demand curve.

In this embodiment, the estimation unit 111 estimates a demand curve using a contextual bandit algorithm. A specific method for estimating a demand curve using a contextual bandit algorithm may be any existing technology, such as the technology disclosed in Patent Document 1, that is used as appropriate.

A demand curve shows the quantity demanded of a product relative to its selling price. Therefore, a fundamental principle in economics is that demand curves slope downward, as shown in demand curve GY in Figure 3. A demand curve slopes downward, meaning that the quantity demanded for a certain selling price is greater than the quantity demanded for a higher selling price, as shown in Figure 3. In the example shown in Figure 3, the quantity demanded D1 for selling price P1 is greater than the quantity demanded D2 for selling price P2 (selling price P1 < selling price P2). Also, in the example shown in Figure 3, the quantity demanded D2 for selling price P2 is greater than the quantity demanded D3 for selling price P3 (selling price P2 < selling price P3). Note that in Figure 3, selling price Pmin is the lower limit of the selling price that can be set for the product (e.g., the purchase price), and selling price Pmax is the upper limit of the selling price that can be set for the product.

As mentioned above, because the contextual bandit algorithm is an algorithm for estimating advertising effectiveness, the demand curve estimated by the estimation unit 111 may not slope downward to the right. For example, the demand curve GYA1 shown in FIG. 4A or the demand curve GYA2 shown in FIG. 4B may be estimated by the estimation unit 111. As shown in FIG. 4A, the demand quantity D1 indicated by the demand curve GYA1 at the selling price P1 is less than both the demand quantity D2 and the demand quantity D3 indicated by the demand curve GY at the selling price P2 and the selling price P3, respectively, and the demand quantity D2 is less than the demand quantity D3. In other words, the demand curve GYA1 slopes upward to the right. The demand curve GYA1 is inappropriate in that it slopes upward to the right. The demand curve GYA2 shown in FIG. 4B slopes downward in relation to the selling price Pmin and the selling price Pmax, but is inappropriate in that the demand quantity D2 is greater than the demand quantity D1.

One aspect of the prior art for estimating a demand curve using a contextual bandit algorithm is to set the upper limit of the confidence interval for the demand quantity at each selling price as the demand quantity at each selling price. A confidence interval for demand quantity refers to a range of demand quantity within which the average actual demand quantity can be trusted to fall with a predetermined probability, such as 90%. Figure 5 shows an aspect in which the upper limit of the confidence interval for the demand quantity at each selling price is set as the demand quantity at each selling price. In Figure 5, the confidence interval for the demand quantity at each selling price is indicated by an arrow. Another aspect of estimating a demand curve using a contextual bandit algorithm is to use Thompson sampling to determine the expected demand quantity at each selling price. The present embodiment employs the former aspect, i.e., to set the upper limit of the confidence interval for the demand quantity at each selling price as the demand quantity at each selling price. In other words, the estimation unit 111 sets the upper limit of the confidence interval for the demand quantity at each selling price as the demand quantity at each selling price. In an embodiment in which the upper limit of the confidence interval for the demand quantity at each selling price is used as the demand quantity at each selling price, the demand quantity represented by the estimated demand curve may be excessively large at any selling price. Note that an embodiment in which the lower limit of the confidence interval for the demand quantity at each selling price is used as the demand quantity at each selling price may also be considered. In an embodiment in which the lower limit of the confidence interval for the demand quantity at each selling price is used as the demand quantity at each selling price, the demand quantity represented by the estimated demand curve may be excessively small at any selling price.

The determination unit 112, acquisition unit 113, and correction unit 114 are responsible for correcting the demand curve estimated by the estimation unit 111.

The determination unit 112 determines a selling price that serves as a reference when correcting the demand curve estimated by the estimation unit 111. Hereinafter, the selling price determined by the determination unit 112 is referred to as the "reference price." In this embodiment, the determination unit 112 determines the reference price based on the width of the confidence interval for the demand quantity calculated for each selling price by the estimation unit 111. More specifically, the determination unit 112 determines the selling price with the narrowest confidence interval as the reference price. Furthermore, the determination unit 112 sets each selling price other than the reference price from the lower limit of the selling price to the upper limit of the selling price as an unprocessed price. The reference price is an example of a first selling price in this disclosure. Each unprocessed price is an example of a second selling price in this disclosure.

The acquisition unit 113 acquires the demand quantity at the base price by referring to the demand curve estimated by the estimation unit 111. The demand quantity at the base price is an example of a first demand quantity in the present disclosure. The acquisition unit 113 also acquires the demand quantity at the raw price by referring to the demand curve estimated by the estimation unit 111. The demand quantity at the raw price is an example of a second demand quantity in the present disclosure.

The correction unit 114 executes a first correction process, a second correction process, and a third correction process. Details will be described later, but the first and second correction processes are processes for correcting a demand curve that is not downward sloping to the right. The third correction process is a process for correcting a demand curve that represents an excessive demand quantity for each selling price.

The first correction process is performed on unprocessed prices that are higher than the reference price. In the first correction process, if the demand volume for the unprocessed price is greater than the demand volume for the reference price, the correction unit 114 corrects the demand curve estimated by the estimation unit 111 to reduce the latter demand volume to or below the former demand volume. If the demand volume for the unprocessed price is less than the demand volume for the reference price, in the first correction process, the correction unit 114 does not correct the demand volume for the unprocessed price. The correction unit 114 changes the unprocessed price for which the first correction process has been performed to the processed price.

The reduction in demand volume in the first correction process is a function whose variable is the difference between the base price and the raw price. The function exhibits a larger value the greater the difference between the base price and the raw price. For example, the increase in demand volume W1 at a first raw price whose difference from the base price is Δ1 is smaller than the decrease in demand volume W2 at a second raw price whose difference from the base price is Δ2 (Δ2 > Δ1). For example, if the demand curve GYA1 shown in Figure 4A is the target of correction and the selling price P1 is the base price PB, the first correction process is performed on the selling prices of the demand curve GYA1 that are higher than the base price PB. As a result of performing the first correction process on the portion higher than the base price PB, the portion of the demand curve GYA1 that is higher than the base price PB is corrected to the demand curve GYB1 shown in Figure 6A.

The second correction process is performed on unprocessed prices that are lower than the base price. In the second correction process, if the demand volume for the base price is greater than the demand volume for the unprocessed price, the correction unit 114 corrects the demand curve estimated by the estimation unit 111 to raise the demand volume for the former to equal or exceed the demand volume for the latter. If the demand volume for the unprocessed price is equal to or greater than the demand volume for the base price, in the second correction process, the correction unit 114 does not correct the demand volume for the unprocessed price. The correction unit 114 changes the unprocessed price for which the second correction process has been performed to the processed price.

The reduction in demand quantity relative to the raw price in the second correction process is also a function whose variable is the difference between the base price and the raw price. This function also exhibits a larger value the greater the difference between the base price and the raw price. For example, if the demand curve GYA1 shown in Figure 4A is the target of correction and the selling price P1 is the base price PB, the second correction process is performed on the selling prices of the demand curve GYA1 that are lower than the base price PB. As a result of performing the second correction process on the portion lower than the base price PB, the portion of the demand curve GYA1 that is lower than the base price PB is corrected to the demand curve GYB2 shown in Figure 6B.

The third correction process corrects the demand curve based on the statistical quantity of the confidence interval width calculated for each selling price by the estimation unit 111. In this embodiment, the statistical quantity is the average value of each confidence interval width, but it may also be the maximum value, minimum value, median, most frequently occurring value, or standard deviation of the confidence interval width. In the third correction process, the correction unit 114 further corrects the demand curve by subtracting a value WM corresponding to the statistical quantity from the demand quantity at each selling price represented by the demand curve corrected by the first and second correction processes. In this embodiment, the value WM is the average value of the confidence interval width × 1/2, but the value WM may be any value calculated by a function using the statistical quantity of the confidence interval width as a variable, such as the square root of the statistical quantity. For example, if the demand curve GYB shown in FIG. 7 is obtained by the first and second correction processes, this demand curve GYB is corrected to the demand curve GYC shown in FIG. 7 by the third correction process.

Furthermore, the processing device 110, operating in accordance with program P, executes the demand curve estimation method shown in FIG. 8 for each store and each product. As shown in FIG. 8, the demand curve estimation method in this embodiment includes the processes of steps SA110 to SA170.

In step SA110, the processing device 110 functions as the estimation unit 111. In step SA110, the processing device 110 acquires sales performance information from the order quantity determination device 20. Next, the processing device 110 estimates a demand curve based on the acquired sales performance information. In addition, in the process of estimating the demand curve, the processing device 110 calculates the width of the confidence interval for each selling price.

In step SA120, which follows step SA110, the processing device 110 functions as the determination unit 112. In step SA120, the processing device 110 determines the base price and unprocessed price based on the width of the confidence interval for the demand volume at each selling price calculated in step SA110.

In step SA130, which follows step SA120, the processing device 110 determines whether there are any unprocessed prices higher than the reference price. If the determination result of step SA130 is "Yes," the processing device 110 executes the processing of step SA140 and then executes the processing of step SA150. In step SA140, the processing device 110 functions as the acquisition unit 113 and the correction unit 114. In step SA140, the processing device 110 executes the first correction processing for any unprocessed prices higher than the reference price. If the determination result of step SA130 is "No," the processing device 110 executes the processing of step SA150 without executing the processing of step SA140.

For example, suppose the demand curve GYA1 shown in Figure 4A is estimated in step SA110. Then, in step SA120, the selling price P1 in Figure 4A is determined to be the base price PB. As shown in Figure 4A, selling price P2, selling price P3, and selling price Pmax are all unprocessed prices higher than selling price P1, so the determination result in step SA130 is "Yes." Therefore, the first correction process is performed on the unprocessed prices of demand curve GYA1 that are higher than selling price P1. As a result, the portion of demand curve GYA1 that is higher than selling price P1 is corrected to the curve GYB1 shown in Figure 6A.

In step SA150, the processing device 110 determines whether there are any unprocessed prices that are lower than the base price. If the determination result in step SA150 is "Yes," the processing device 110 executes the processing of step SA160 and then executes the processing of step SA170. In step SA160, the processing device 110 functions as the acquisition unit 113 and the correction unit 114. In step SA160, the processing device 110 executes the second correction processing for unprocessed prices that are lower than the base price PB. If the determination result in step SA150 is "No," the processing device 110 executes the processing of step SA170 without executing the processing of step SA160.

When the demand curve GYA1 shown in Figure 4A is estimated in step SA110 and the selling price P1 is the base price PB, there is a selling price Pmin that is an unprocessed price lower than the base price PB. Therefore, the determination result of step SA150 is "Yes," and processing of step SA160 is executed. In step SA160, a second correction process is executed for the portion of the demand curve GYA1 that is lower than the base price PB. As a result, the portion of the demand curve GYA1 that is lower than the base price PB is corrected to the curve GYB2 shown in Figure 6B.

At the time step SA170 is executed, the demand curve estimated in step SA110 has been corrected to a downward-sloping demand curve. In step SA170, processing device 110 functions as correction unit 114. In step SA170, processing device 110 performs a third correction process on the demand curve corrected to a downward-sloping demand curve. By performing the process of step SA170, the demand quantity at each selling price represented by the demand curve is reduced by an amount corresponding to the statistical value of the width of the confidence interval. For example, if the demand curve at the time step SA170 is executed is demand curve GYB in Figure 7, demand curve GYB is corrected to demand curve GYC shown in Figure 7 by the third correction process.

The demand curve estimation device 10 of this embodiment generates a downward-sloping demand curve. Furthermore, the demand quantity at each selling price represented by the demand curve generated by the demand curve estimation device 10 is not excessively large.

Figure 9 is a flowchart showing the flow of the order quantity determination method executed in the order support system 1. As shown in Figure 9, the order quantity determination method includes the processes of steps SB110 and SB120. Step SB110 is a process executed by the demand curve estimation device 10. In step SB110, the demand curve estimation device 10 executes the demand curve estimation method shown in Figure 8 for each product at each of stores 40A and 40B, thereby generating a demand curve for each store and each product. Step SB120 is a process executed by the order quantity determination device 20. In step SB120, the order quantity determination device 20 determines the order quantity and sales price of each product that maximizes gross profit for each store, based on the inventory amount of the product at the store and the demand curve generated in step SB120.

As described above, according to this embodiment, the demand curve is estimated using a contextual bandit algorithm, so there is no problem with estimating the demand curve even in situations where the sales history of a product is not sufficiently accumulated. Furthermore, even if the demand curve estimated using the contextual bandit algorithm is an inappropriate demand curve, it is corrected to an appropriate demand curve by the demand curve estimation device 10. In the ordering support system 1, the order quantity of a product is determined based on an appropriate demand curve, so the occurrence of excess inventory is avoided. Because the occurrence of excess inventory is avoided, gross profit can be maximized.

<B. Transformation>
One embodiment of the present disclosure has been described above, but this embodiment may be modified as follows.
(1) In the above embodiment, the selling price with the narrowest confidence interval was determined as the base price, but the base price may be the lower limit or upper limit of the selling price of the product. In an aspect in which the lower limit or upper limit of the selling price of the product is set as the base price, the determination unit 112 may be omitted. Furthermore, in an aspect in which the determination unit 112 is omitted, the first correction process may be performed by sequentially selecting base prices and unprocessed prices in ascending order from the lower limit of the selling price until the upper limit of the selling price is selected as the unprocessed selling price. In this aspect, the second correction process may be omitted. Similarly, the second correction process may be performed by sequentially selecting base prices and unprocessed prices in descending order from the upper limit of the selling price until the lower limit of the selling price is selected as the unprocessed selling price. In this aspect, the first correction process may be omitted.

(2) In the above embodiment, the estimation unit 111 estimated the demand curve using a contextual bandit algorithm, but the demand curve may be estimated using other algorithms. Furthermore, in the above embodiment, the correction unit 114 performed the first correction process, the second correction process, and the third correction process. However, the third correction process may be omitted and only the first correction process and the second correction process may be performed, or conversely, only the third correction process may be performed. In an aspect in which the correction unit 114 performs only the third correction process, the determination unit 112 and the acquisition unit 113 may be omitted, and the correction unit 114 may simply perform the third correction process on the demand curve estimated by the estimation unit 111.

(3) In the above embodiment, the third correction process subtracts a value corresponding to the statistical quantity of the width of the confidence interval calculated for each selling price by the estimation unit 111 from the demand quantity at each selling price represented by the demand curve corrected by the first correction process and the second correction process. However, the estimation unit 111 may generate a demand curve based on the lower limit of the confidence interval at each selling price. In this aspect, the third correction process may add a value corresponding to the statistical quantity of the width of the confidence interval to the demand quantity at each selling price represented by the demand curve corrected by the first correction process and the second correction process. According to this aspect, an inappropriate demand curve that represents an insufficient demand quantity for each selling price is corrected to a demand curve that represents an appropriate demand quantity, thereby avoiding opportunity losses due to stockouts.

(4) In the above embodiment, stores 40A and 40B were retail stores such as convenience stores, but they may also be micro-stores that sell merchandise unattended. A micro-store is a device that includes shelves on which merchandise is displayed, a detection device, a communication device, and a payment device. The detection device in a micro-store detects the type and number of merchandise removed from the shelves by a user. The communication device is connected to the communication network 30. The communication device transmits sales performance information corresponding to the detection results by the detection device to the order quantity determination device 20. The payment device is a device that performs payment processing according to the detection results by the detection device. Furthermore, while the merchandise in the above embodiment was a product, it may also be a service.

<C: Other>
(1) In the above embodiment and variations, the non-volatile storage device 132 may include a flexible disk, a magneto-optical disk (e.g., a compact disk, a digital versatile disk, a Blu-ray (registered trademark) disk), a smart card, a flash memory device (e.g., a card, a stick, a key drive), a CD-ROM (Compact Disc-ROM), a register, a removable disk, a hard disk, a floppy (registered trademark) disk, a magnetic strip, a database, a server, or other suitable storage medium. The program may also be transmitted from a network via a telecommunications line.

(2) Each of the above embodiments and variations may be applied to systems using LTE (Long Term Evolution), LTE-A (LTE-Advanced), SUPER 3G, IMT-Advanced, 4G, 5G, FRA (Future Radio Access), W-CDMA (registered trademark), GSM (registered trademark), CDMA2000, UMB (Ultra Mobile Broadband), IEEE 802.11 (Wi-Fi), IEEE 802.16 (WiMAX), IEEE 802.20, UWB (Ultra-Wideband), Bluetooth (registered trademark), or other suitable systems, and/or next-generation systems based on and enhanced by these.

(3) The information described in each of the above embodiments and modifications may be represented using any of a variety of different technologies. For example, data, information, etc. that may be referred to throughout the above description may be represented by voltage, current, electromagnetic waves, magnetic fields, magnetic particles, optical fields, photons, or any combination thereof.
It should be noted that terms explained in this specification and/or terms necessary for understanding this specification may be replaced with terms having the same or similar meanings.

(4) In each of the above embodiments and variations, input and output information may be stored in a specific location (e.g., the volatile storage device 131) or managed using a management table. Input and output information may be overwritten, updated, or added to. Output information may be deleted. Input information may be sent to another device.

(5) In each of the above embodiments and variations, the determination may be based on a value represented by one bit (0 or 1), a Boolean value (true or false), or a comparison of numerical values (e.g., a comparison with a predetermined value).

(6) The order of the process procedures, sequences, flowcharts, etc. illustrated in each of the above embodiments and variations may be changed as long as there is no contradiction. For example, the methods described herein present various step elements in an exemplary order and are not limited to the specific order presented.

(7) Each function illustrated in FIG. 2 is realized by any combination of hardware and software. Furthermore, the method for realizing each function is not particularly limited. That is, each function may be realized using a single device that is physically or logically coupled, or may be realized using two or more devices that are physically or logically separated and connected directly or indirectly (e.g., using a wired or wireless connection) and these multiple devices. The functional blocks may be realized by combining software with the single device or the multiple devices.
Functions include, but are not limited to, judgment, determination, judgment, calculation, computation, processing, derivation, investigation, search, confirmation, reception, transmission, output, access, resolution, selection, selection, establishment, comparison, assumption, expectation, regard, broadcasting, notifying, communicating, forwarding, configuring, reconfiguring, allocating, mapping, and assignment. For example, a functional block (component) that performs transmission is called a transmitting unit or transmitter. As mentioned above, there are no particular limitations on how each is implemented.

(8) The programs exemplified in each of the above embodiments and variants should be broadly interpreted to mean instructions, instruction sets, code, code segments, program code, subprograms, software modules, applications, software applications, software packages, routines, subroutines, objects, executable files, threads of execution, procedures, or functions, etc., regardless of whether they are called software, firmware, middleware, microcode, hardware description language, or by other names.
Software, instructions, etc. may also be transmitted or received over a transmission medium. For example, if the software is transmitted from a website, server, or other remote source using wired technologies such as coaxial cable, fiber optic cable, twisted pair, and Digital Subscriber Line (DSL), and/or wireless technologies such as infrared, radio, and microwave, these wired and/or wireless technologies are included within the definition of transmission media.

(9) The demand curve estimation device 10 may be a mobile station. A mobile station may also be referred to by those skilled in the art as a subscriber station, mobile unit, subscriber unit, wireless unit, remote unit, mobile device, wireless device, wireless communication device, remote device, mobile subscriber station, access terminal, mobile terminal, wireless terminal, remote terminal, handset, user agent, mobile client, client, or some other suitable terminology.

(10) In each of the above embodiments and variations, the phrase "based on" does not mean "based only on," unless otherwise specified. In other words, the phrase "based on" means both "based only on" and "based at least on."

(11) In each of the above embodiments and variations, the terms "connected," "coupled," or any variation thereof, refer to any direct or indirect connection or coupling between two or more elements, and may include the presence of one or more intermediate elements between two elements that are "connected" or "coupled" to each other. The coupling or connection between elements may be physical, logical, or a combination thereof. For example, "connected" may be read as "access." As used in this disclosure, two elements may be considered to be "connected" or "coupled" to each other using at least one of one or more wires, cables, and printed electrical connections, as well as electromagnetic energy having wavelengths in the radio frequency range, microwave range, and optical (both visible and invisible) range, as some non-limiting and non-exhaustive examples.

(12) In each of the above embodiments and variations, to the extent that the terms "including," "comprising," and variations thereof are used in this specification or the claims, these terms are intended to be inclusive, similar to the term "comprise." Furthermore, the term "or" used in this specification or the claims is not intended to be an exclusive or.

(13) Throughout this application, where articles have been added by translation, such as a, an, and the in English, these articles include the plural unless the context clearly indicates otherwise.

(14) In this specification, the term "section" may be replaced with other terms such as circuit, device, or unit. Similarly, the term "apparatus" may be replaced with other terms such as circuit, device, or unit.

(15) As used in this disclosure, the terms "system" and "network" are used interchangeably.

(16) As used in this disclosure, the terms "determining" and "determining" may encompass a wide variety of actions. "Determining" and "determining" may include, for example, judging, calculating, computing, processing, deriving, investigating, looking up, searching, inquiring (e.g., searching a table, database, or other data structure), and ascertaining something that is considered to be a "determining." Also, "determining" and "determining" may include receiving (e.g., receiving information), transmitting (e.g., sending information), input, output, accessing (e.g., accessing data in memory), and so on. Furthermore, "judgment" and "decision" can include regarding actions such as resolving, selecting, choosing, establishing, and comparing as having been "judgment" or "decision." In other words, "judgment" and "decision" can include regarding some action as having been "judgment" or "decision." Furthermore, "judgment (decision)" can be interpreted as "assuming," "expecting," "considering," etc.

(17) It will be apparent to those skilled in the art that the present disclosure is not limited to the embodiments described herein. The present disclosure can be implemented in modified and altered forms without departing from the spirit and scope of the present disclosure as defined by the claims. Therefore, the descriptions in this specification are for illustrative purposes only and do not have any limiting meaning on the present disclosure. Furthermore, multiple aspects selected from the aspects exemplified in this specification may be combined.

<D: Aspects understood from the above-described embodiments or modifications>
The following aspects can be understood from at least one of the above-described embodiments or modifications.
A demand curve estimation device according to a first aspect includes an estimation unit, a first acquisition unit, and a correction unit. The estimation unit estimates a demand curve indicating an expected demand quantity for a commodity, which is a good or service, based on a sales history indicating the sales volume of the commodity at each of a plurality of different prices. The first acquisition unit acquires a first demand quantity calculated from the demand curve for a first selling price and a second demand quantity calculated from the demand curve for a second selling price different from the first selling price. When the second selling price is lower than the first selling price and the first demand quantity is greater than the second demand quantity, the correction unit corrects the demand curve to increase the second demand quantity to or above the first demand quantity. When the first selling price is lower than the second selling price and the second demand quantity is greater than the first demand quantity, the correction unit corrects the demand curve to decrease the second demand quantity to or below the first demand quantity. According to the demand curve estimation device of the first aspect, the demand curve estimated from sales performance is corrected to a demand curve indicating that the higher the sales price, the lower the demand quantity, i.e., a demand curve that slopes downward to the right.

In an example of the first aspect (second aspect), the estimation unit may estimate the demand curve using a contextual bandit algorithm. According to the demand curve estimation device of the second aspect, the demand curve estimated using the contextual bandit algorithm is corrected to a downward-sloping demand curve.

The demand curve estimation device in the second example (third example) may include a determination unit that determines the first selling price based on the width of a confidence interval for the demand quantity of the commodity calculated by the contextual bandit algorithm. According to the demand curve estimation device of the third example, the first selling price that serves as a reference point when correcting the demand curve estimated according to the contextual bandit algorithm based on sales performance is selected based on the width of the confidence interval.

In an example of the third aspect (fourth aspect), the determination unit may determine the selling price with the narrowest confidence interval as the first selling price. According to the demand curve estimation device of the fourth aspect, the price with the narrowest confidence interval, i.e., the price estimated to have the highest accuracy in demand volume, is selected as the first selling price, thereby improving the accuracy of demand volume at each selling price represented by the corrected demand curve.

The demand curve estimation device according to the fifth aspect comprises an estimation unit and a correction unit. The estimation unit estimates a demand curve indicating the expected demand quantity of a commodity, which is a product or service, relative to the selling price of the commodity using a contextual bandit algorithm based on sales records indicating the sales volume of the commodity at each of a plurality of different prices. The correction unit corrects the demand quantity at each selling price indicated by the demand curve in accordance with statistics of the width of the confidence interval for the demand quantity of the commodity calculated for each selling price by the contextual bandit algorithm. The demand curve estimation device according to the fifth aspect can reduce the risk of the demand quantity for each selling price indicated by the demand curve estimated using the contextual bandit algorithm being too high or too low.

In an example of the fifth aspect (sixth aspect), the correction unit may add a value corresponding to the statistical quantity to the demand quantity at each selling price indicated by the demand curve, or subtract the value from the demand quantity at each selling price indicated by the demand curve. According to the demand curve estimation device of the sixth aspect, the demand curve estimated using the contextual bandit algorithm is translated along the coordinate axis indicating the demand quantity.

The ordering support system according to the seventh aspect includes a demand curve estimation device according to any one of the first to sixth aspects and an order quantity estimation device. The order quantity estimation device acquires the inventory quantity of the merchandise in the store and determines the order quantity of the merchandise based on the acquired inventory quantity and a demand curve corrected using the demand curve estimation device.

1...Order support system, 10...Demand curve estimation device, 20...Order quantity determination device, 30...Communication network, 40A, 40B...Store, 110...Processing device, 111...Estimation unit, 112...Determination unit, 113...Acquisition unit, 114...Correction unit, 120...Communication device, 130...Storage device, 131...Volatile storage device, 132...Non-volatile storage device, P...Program.

Claims (3)

an estimation unit that estimates a demand curve showing the expected demand for a commodity, which is a product or service, relative to the sales price of the commodity, using a contextual bandit algorithm based on sales records showing the sales quantities of the commodity at each of a plurality of different prices;
a determination unit that determines a selling price with the narrowest confidence interval for the demand quantity of the commodity calculated by the contextual bandit algorithm as a first selling price;
an acquisition unit that acquires a first demand amount calculated from the demand curve for the first selling price and a second demand amount calculated from the demand curve for a second selling price different from the first selling price;
a correction unit that corrects the demand curve to raise the second demand amount to or above the first demand amount when the second selling price is lower than the first selling price and the first demand amount is greater than the second demand amount, and corrects the demand curve to lower the second demand amount to or below the first demand amount when the first selling price is lower than the second selling price and the second demand amount is greater than the first demand amount;
A demand curve estimating device having the above. an estimation unit that estimates a demand curve showing the expected demand for a commodity, which is a product or service, relative to the sales price of the commodity, using a contextual bandit algorithm based on sales records showing the sales quantities of the commodity at each of a plurality of different prices;
a correction unit that corrects the demand quantity at each selling price indicated by the demand curve in accordance with statistics of the width of a confidence interval for the demand quantity of the commodity calculated for each selling price of the commodity by the contextual bandit algorithm;
The correction unit
When the demand quantity calculated by the contextual bandit algorithm is the lower limit of the confidence interval, a value corresponding to the statistical quantity is added to the demand quantity at each selling price indicated by the demand curve, and when the demand quantity calculated by the contextual bandit algorithm is the upper limit of the confidence interval, a value corresponding to the statistical quantity is subtracted from the demand quantity at each selling price indicated by the demand curve.
Demand curve estimator. The demand curve estimation device according to claim 1 or 2;
an order quantity determination device that acquires an inventory amount of the merchandise in the store and determines an order quantity of the merchandise based on the acquired inventory amount and the demand curve corrected using the demand curve estimation device;
An ordering support system equipped with: