Information processing device and information processing method

Abstract

A calculation unit 14 of an information processing device 10 calculates effect information pertaining to the effect of a new campaign by using a hierarchical Bayesian model in which stores that conducted a plurality of campaigns in the past and the business types of the stores are included in different hierarchical levels, and acquired new campaign information. In the hierarchical Bayesian model, when yi is defined as an estimated value of the causal effect of each campaign, θi is defined as an average, and σi 2 is defined as a variance, yi~N (θi, σi 2). When a Gaussian process is set with zai defined as a j-dimensional store dummy variable, zbi defined as an i-dimensional business type dummy variable, xi defined as other explanatory variables, α defined as an intercept, θa defined as a store effect, θb defined as a business type effect, β defined as a coefficient of xi, τa defined as the variation of θa, τb defined as the variation of θb, zci defined as a period dummy variable, and θc defined as a prior distribution for an effect corresponding to a period when the campaign was conducted, θi=α+θa Tzai+θb Tzbi +βTxi+θc Tzci (where, θa T,zai, θb T,zbi, βT,xi, θc T, and zci each are a vector), θ aj~(0,τa 2), j=1,…,J, θbk~(0,τb 2), k=1,…,K, and θ c~N(0N,C(σ2 p,lp)).

Description

Information processing device and information processing method

[0001] This invention relates to a technology for predicting the effectiveness of campaigns conducted in stores and other similar establishments.

[0002] Various measures (hereinafter referred to as "campaigns") are being implemented for the purpose of sales promotion. For example, Patent Document 1 discloses a method for identifying customer groups who purchased the target product during the campaign period and other customer groups, and calculating the effectiveness of the campaign from the difference in the sales growth rate during the promotion period among these identified customer groups.

[0003] Japanese Patent Publication No. 2013-57990

[0004] In light of the above background, the present invention aims to predict the effectiveness of a new campaign by utilizing knowledge gained from past campaigns.

[0005] To solve the above problems, the present invention provides an information processing device comprising: an acquisition unit that acquires new campaign information including information about stores that will implement a new campaign and the type of business to which each store belongs; a hierarchical Bayesian model that includes stores that have implemented multiple campaigns in the past and the types of business to which each store belongs at different hierarchical levels; a variable that represents the effect according to the timing of when the campaign was implemented; a calculation unit that uses the acquired new campaign information to calculate effect information regarding the effect of the new campaign; and an output unit that outputs the calculated effect information.

[0006] Furthermore, the present invention provides an information processing method characterized by comprising: an acquisition step of acquiring new campaign information including information on stores that will implement a new campaign and the type of business to which each store belongs; a calculation step of calculating effect information regarding the effect of the new campaign using a hierarchical Bayesian model that includes stores that have implemented multiple campaigns in the past and the types of business to which each store belongs at different hierarchical levels; a variable that represents the effect according to the timing of when the campaign was implemented; and an output step of outputting the calculated effect information.

[0007] According to the present invention, the effectiveness of a new campaign can be estimated by utilizing knowledge gained from past campaigns.

[0008] This is a block diagram showing an example of the hardware configuration of an information processing device 10 according to one embodiment of the present invention. This is a block diagram showing an example of the functional configuration of the information processing device 10. This is a diagram illustrating the overview of the hierarchical Bayesian model in this embodiment. This is a flowchart showing an example of the operation of the information processing device 10. This is a diagram illustrating a horseshoe prior.

[0009] [Embodiment] [Configuration] Figure 1 is a diagram showing the hardware configuration of an information processing device 10 according to an embodiment of the present invention. Physically, the information processing device 10 is configured as a computer including a processor 1001, memory 1002, storage 1003, communication device 1004, input device 1005, output device 1006, and a bus connecting these. Each of these devices operates on power supplied from a battery (not shown). In the following description, the word "device" can be read as a circuit, device, unit, etc. The hardware configuration of the information processing device 10 may be configured to include one or more of the devices shown in Figure 1, or it may be configured to omit some of the devices. Alternatively, multiple devices with different housings may be connected to each other via communication to constitute the information processing device 10.

[0010] Each function in the information processing device 10 is realized by loading predetermined software (programs) onto hardware such as the processor 1001 and memory 1002, which allows the processor 1001 to perform calculations, control communication by the communication device 1004, and control at least one of data reading and writing in the memory 1002 and storage 1003.

[0011] The processor 1001 controls the entire computer, for example, by running an operating system. The processor 1001 may be composed of a central processing unit (CPU) that includes interfaces with peripheral devices, control devices, arithmetic units, registers, etc. Also, for example, a baseband signal processing unit or a call processing unit may be implemented by the processor 1001.

[0012] The processor 1001 reads programs (program code), software modules, data, etc., from at least one of the storage 1003 and the communication device 1004 into the memory 1002 and executes various processes accordingly. The program used is one that causes the computer to execute at least a part of the operations described later. Functional blocks of the information processing device 10 may be stored in the memory 1002 and implemented by control programs that run on the processor 1001. Various processes may be executed by one processor 1001, but may also be executed simultaneously or sequentially by two or more processors 1001. The processor 1001 may be implemented by one or more chips. The program may also be transmitted to the information processing device 10 via a telecommunications line.

[0013] The memory 1002 is a computer-readable recording medium and may consist of at least one of the following: ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable ROM), RAM (Random Access Memory), etc. The memory 1002 may also be called a register, cache, main memory, etc. The memory 1002 can store executable programs (program code), software modules, etc., for carrying out the method according to this embodiment.

[0014] The storage 1003 is a computer-readable recording medium and may consist of at least one of the following: an optical disc such as a CD-ROM (Compact Disc ROM), a hard disk drive, a flexible disk, a magneto-optical disk (e.g., a compact disc, a digital multipurpose disc, a Blu-ray® disc), a smart card, flash memory (e.g., a card, a stick, a key drive), a floppy® disk, a magnetic strip, etc. The storage 1003 may also be called an auxiliary storage device.

[0015] The communication device 1004 is hardware (transceiver / receiver device) for communicating between computers via at least one of a wired network and a wireless network, and is also referred to as a network device, network controller, network card, communication module, etc.

[0016] Each device, such as the processor 1001 and the memory 1002, is connected by a bus for communicating information. The bus may be configured using a single bus, or different buses may be used for each device.

[0017] The information processing device 10 may include hardware such as a microprocessor, a digital signal processor (DSP), an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), and an FPGA (Field Programmable Gate Array), and some or all of each functional block may be realized by this hardware. For example, the processor 1001 may be implemented using at least one of these hardware components.

[0018] Figure 2 is a block diagram showing the functional configuration of the information processing device 10. In the information processing device 10, the processor 1001 reads programs and the like from the storage 1003 into the memory 1002 and executes them, thereby realizing the functions of the acquisition unit 11, the storage unit 12, the generation unit 13, the calculation unit 14, and the output unit 15.

[0019] The acquisition unit 11 acquires information from an external source. For example, the acquisition unit 11 acquires past campaign information, which is information about multiple campaigns that have been conducted in the past. Past campaign information includes, for example, information about the stores that implemented each past campaign and the type of business to which those stores belong, as well as information about the effectiveness of each campaign. Information about the stores that implemented each past campaign includes, for example, the name of the store (or the name of the franchise chain if the store is a franchise chain store). Information about the type of business to which the stores that implemented each past campaign belong is a classification item based on differences in the operating style of stores in the retail and food service industries. For example, in the retail industry, this could include supermarkets, convenience stores, department stores, fast food restaurants, drugstores, consumer electronics stores, fashion stores, discount stores, etc. In this embodiment, "store" may include not only stores that exist in real three-dimensional space, but also, for example, online shops on the internet. Information about the effectiveness of each campaign is information that shows the effect when comparing before and after each campaign, such as sales, profits, and customer satisfaction obtained from that campaign. Information demonstrating these effects can be obtained by those who conduct the campaign, by calculating the difference in sales and profits before and after the campaign, or by analyzing changes in customer satisfaction based on surveys, etc.

[0020] Furthermore, the acquisition unit 11 acquires new campaign information, which is information about new campaigns that are scheduled to be implemented. New campaign information includes, for example, information about the stores that will implement the new campaign and the type of business to which each store belongs. Information about the stores that will implement the new campaign includes, for example, the name of the store (or the name of the franchise chain if the store is a franchise chain store). Information about the type of business to which the stores that will implement the new campaign belong includes classification items based on differences in business format within the retail and food service industries. For example, in the retail industry, these may include supermarkets, convenience stores, department stores, fast food restaurants, drugstores, consumer electronics stores, fashion stores, discount stores, etc.

[0021] The information acquired by the acquisition unit 11 as described above is stored in the storage unit 12.

[0022] The generation unit 13 generates a hierarchical Bayesian model using the past campaign information acquired by the acquisition unit 11, that is, information on the stores where a plurality of past campaigns were each implemented and the business types to which those stores belong, and information on the effects of each of those campaigns. A hierarchical Bayesian model is a statistical model in a hierarchical form described in a plurality of different hierarchies. Hierarchical Bayes is based on a Bayesian model that quantifies uncertain posterior analysis, and models considering the trends of individual data and the trends of overall data among data with different properties. As shown in FIG. 3, the hierarchical Bayesian model of the present embodiment has a hierarchical structure with multiple hierarchies, where the topmost layer is the "business type" to which the store belongs, the next layer is the "store" where the campaign was implemented, and the bottommost layer is the "campaign" that was held, in the entire store group.

[0023] The details of the hierarchical Bayesian model according to the present embodiment are as follows. First, let y i , i , 2 , i , ai , bi , 2 , , i ,

[0024] , i , i be the estimated value of the causal effect of each campaign, and let θ i be the mean, and let σ i 2 be the variance. Then, y i ~N(θ i , σ i 2 ). That is, each observed value (the estimated value of the causal effect of each campaign y i ) in the hierarchical Bayesian model follows a normal distribution with mean θ <000001​​​​​​​​​​​​​​​a Store effect, θ b Use the business model effect and β as x i The coefficient of τ a θ a Variation, τ b θ b Variation, θ c The time effect of campaign i, according to when it was conducted, z ci When θ is used as a dummy variable representing the timing of campaign i, i = α + θ a T z ai +θ b T z bi +β T x i +θ c T z ci (Note, θ a T ,z ai ,θ b T ,z bi ,β T ,x i ,θ c T ,z ci (Each is a vector) θ aj ~(0,τ a 2 ),j=1,…,J θ bk ~(0,τ b 2 ),k=1,…,K θ c ~N(0 N ,C(σ 2 p ,l p Let's define this formula as follows: This formula divides the effect of each campaign into fixed effects (α,β) which are common effects among these campaigns, and variable effects (θ) which are random effects (store effects, business type effects) that occur between campaigns. a , θ b ) and the variable θ, which is a random effect (time effect) depending on when the campaign was conducted. c This is formulated as a mixed model that combines two elements. For the variable θc corresponding to the time effect, a Gaussian process is set as the prior distribution.

[0025] In the above formula, T represents the transpose matrix. θ aj , θ bk These are independent normal distributions, and are generated and updated independently during the creation (i.e., training) of the hierarchical Bayesian model. a , θ b We assume that they do not influence each other, that is, they are independent. Store dummy variable z ai There are as many dummy variables as there are stores, j in total. One variable takes the value "1" when a store is found and the other takes the value "0" when it is not found. For example, when the store name is "Franchise Chain XXX", the dummy variable z represents the stores that belong to Franchise Chain XXX. ai The variable z represents stores other than those in the franchise chain XXX. ai All values ​​are "0". Note that the store dummy variable z ai If there are multiple stores, then z for each of these stores. ai = 1. Similarly, the business type dummy variable z bi There are as many variables as there are business types, i in number. It takes the value "1" when a business type is matched and the value "0" when it is not matched. For example, when the business type is "convenience store", there is a business type dummy variable z corresponding to convenience stores. bi z is a dummy variable for business types other than convenience stores, where z is "1". bi All values ​​are "0". Note that the business type dummy variable z bi If there are multiple applicable business types, then z for each of these multiple business types. bi = 1. The dummy variable z is a time variable. ci This variable corresponds to the implementation period of each campaign, such as January 2023, February 2023, ..., December 2023.

[0026] The effectiveness of a campaign is thought to vary depending on the season, or time of year, in which the campaign is conducted. c This refers to seasonality, or effects that correspond to the time of year. C(σ 2 p ,l p In the term, θ is the variance-covariance matrix induced by an N×N Gaussian process. cTreat ui as a function of ti. The kernel function K is represented by the following equation. K(t, t') = σ 2 p exp(−2sin(π|t - t'| / P e ) / l 2 p ) p e , l p , σ p are parameters that characterize the kernel function K. For example, when expressing the periodicity of each month by focusing on the profitability in annual units (considering leap years), pe = 12. Also, α ∼ N(m a , s 2 α ), μ a ∼ N(0, s 2 a ), μ b ∼ N(0, s 2 b ) β ∼ N(M β , S β ), σ 2 p ∼ HC(δ σ ), l p ∼ HC(δ l ) τ 2 a ∼ HC(δ a ), τ 2 b ∼ HC(δ b ) That is. Note that HC(δ) is a half-Cauchy prior with scale parameter δ.

[0027] By the hierarchical Bayesian model represented by the above formula, τ a , τ b , θ i can be estimated.

[0028] Note that the store dummy variable z ai and the business format dummy variable z bi are assumed to be sparse, that is, the number of data is not sufficient for the required accuracy when the number of campaigns for each store and business format is not so large. Therefore, the store effect θ a and the business format effect θ bThe prior distribution may be a horseshoe prior, as illustrated in Figure 4. This allows for addressing the sparsity of the data, such as when the number of campaigns for each store and business type is not very large. The hierarchical Bayesian model generated by the generation unit 13 is stored in the storage unit 12.

[0029] In Figure 2, the calculation unit 14 uses the new campaign information acquired by the acquisition unit 11—that is, information about the stores implementing the new campaign and the type of business each store belongs to—and the hierarchical Bayesian model stored in the storage unit 12 to calculate effect information regarding the effectiveness of the new campaign. Specifically, the calculation unit 14 calculates the effect information of the new campaign by inputting the new campaign information into the hierarchical Bayesian model. This effect information is expressed in the same format as the information on the campaign's effectiveness used when generating the hierarchical Bayesian model, and includes, for example, sales, profits, and customer satisfaction obtained from the new campaign.

[0030] The output unit 15 outputs the effect information calculated by the calculation unit 14 to the outside. Output here includes all forms of output, such as transmission, display, writing to a storage medium, and printout.

[0031] [Operation] Next, the operation of this embodiment will be described. Figure 5 is a flowchart showing an example of the operation of the information processing device 10. Before the process shown in Figure 5 is started, it is assumed that the hierarchical Bayesian model generated by the generation unit 13 is stored in the storage unit 12.

[0032] The acquisition unit 11 acquires new campaign information that has been input to the information processing device 10, for example, via a communication network (not shown) (step S11). The new campaign information acquired by the acquisition unit 11 is stored in the storage unit 12.

[0033] The calculation unit 14 uses the new campaign information acquired by the acquisition unit 11 and stored in the storage unit 12, and the hierarchical Bayesian model stored in the storage unit 12 to calculate effect information regarding the effect of the new campaign (step S12).

[0034] The output unit 15 outputs the effect information calculated by the calculation unit 14 to the outside (step S13).

[0035] According to the above embodiment, it becomes possible to predict the effectiveness of a new campaign with sufficient accuracy by utilizing insights from past campaigns.

[0036] [Modifications] The present invention is not limited to the embodiments described above. The embodiments described above may be modified as follows. Furthermore, two or more of the following modifications may be combined and implemented.

[0037] The block diagram used in the description of the above embodiment shows functional units. These functional blocks (components) are realized by any combination of at least one of hardware and software. Furthermore, the method of realizing each functional block is not particularly limited. That is, each functional block may be realized using one device that is physically or logically coupled, or it may be realized using two or more physically or logically separated devices that are directly or indirectly connected (for example, using wired or wireless connections). A functional block may also be realized by combining the one or more devices with software.

[0038] Functions include, but are not limited to, judgment, decision, determination, calculation, calculation, processing, derivation, investigation, exploration, confirmation, reception, transmission, output, access, resolution, selection, selection, establishment, comparison, assumption, expectation, assumption, broadcasting, notifying, communicating, forwarding, configuring, reconfiguring, allocating (mapping), and assigning. For example, a functional block (configuration part) that enables transmission is called a transmission unit or transmitter. In all cases, as mentioned above, the method of implementation is not particularly limited.

[0039] For example, the information processing device 10 in one embodiment of the present disclosure may function as a computer that performs the processing of the present disclosure.

[0040] Each aspect / embodiment described in this disclosure may be applied to at least one of the following systems: LTE (Long Term Evolution), LTE-A (LTE-Advanced), SUPER 3G, IMT-Advanced, 4G (4th generation mobile communication system), 5G (5th generation mobile communication system), FRA (Future Radio Access), NR (new Radio), W-CDMA®, GSM®, CDMA1000, UMB (ULTRA Mobile Broadband), IEEE 802.11 (Wi-Fi®), IEEE 802.16 (WiMAX®), IEEE 802.20, UWB (ULTRA-WIDEBAND), Bluetooth®, and other appropriate systems, as well as next-generation systems extended based thereon. Furthermore, multiple systems may be applied in combination (for example, a combination of at least one of LTE and LTE-A with 5G).

[0041] The present invention may also be an information processing method characterized by comprising: an acquisition step of acquiring campaign information including information about stores that will implement a new campaign to be conducted and the type of business to which each store belongs; a hierarchical Bayesian model that includes stores that have previously implemented multiple campaigns and the types of business to which each store belongs at different hierarchical levels; a calculation step of calculating effect information regarding the effect of the new campaign using the acquired campaign information; and an output step of outputting the calculated effect information. The processing procedures, sequences, flowcharts, etc., of each aspect / embodiment described in this disclosure may be rearranged in order as long as they do not contradict each other. For example, in the method described in this disclosure, elements of various steps are presented using an exemplary order and are not limited to the specific order presented.

[0042] Input and output information may be stored in a specific location (e.g., memory) or managed using a management table. Input and output information may be overwritten, updated, or appended to. Output information may be deleted. Input information may be transmitted to other devices.

[0043] The determination may be made by a value represented by one bit (0 or 1), by a boolean value (true or false), or by a numerical comparison (for example, a comparison with a predetermined value).

[0044] Although the present disclosure has been described in detail above, it will be clear 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 intent and scope of the present disclosure as defined by the claims. Therefore, the descriptions in the present disclosure are illustrative and not intended to be restrictive in any way.

[0045] Software should be broadly interpreted to mean instructions, instruction sets, code, code segments, program code, programs, subprograms, software modules, applications, software applications, software packages, routines, subroutines, objects, executable files, execution threads, procedures, functions, etc., whether they are called software, firmware, middleware, microcode, hardware description languages, or by any other name. Furthermore, software, instructions, information, etc., may be transmitted and received via a transmission medium. For example, if software is transmitted from a website, server, or other remote source using at least one of wired technologies (such as coaxial cable, fiber optic cable, twisted pair, or digital subscriber line (DSL)) and wireless technologies (such as infrared or microwave), at least one of these wired and wireless technologies is included in the definition of a transmission medium.

[0046] The information, signals, etc., described herein may be represented using any of the following different technologies. For example, data, instructions, commands, information, signals, bits, symbols, chips, etc., which may be referred to throughout the above description, may be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, optical fields or photons, or any combination thereof. Terms used herein and terms necessary for understanding this disclosure may be replaced with terms having the same or similar meaning.

[0047] Furthermore, the information, parameters, etc., described in this disclosure may be expressed using absolute values, relative values ​​from a predetermined value, or corresponding other information.

[0048] In this disclosure, the phrase "based on" does not mean "based solely on" unless otherwise specified. In other words, the phrase "based on" means both "based solely on" and "based at least on."

[0049] Any reference to elements using the designations “First,” “Second,” etc., as used in this disclosure does not generally limit the quantity or order of those elements. These designations may be used in this disclosure as a convenient way to distinguish between two or more elements. Accordingly, references to the First and Second elements do not imply that only two elements may be employed, or that the First element must precede the Second element in any way.

[0050] In the above-described configuration of each device, the term "part" may be replaced with "means," "circuit," "device," etc.

[0051] Where the terms “include,” “including,” and variations thereof are used in this disclosure, these terms are intended to be inclusive, as is the term “comprising.” Furthermore, the term “or” as used in this disclosure is not intended to mean exclusive OR.

[0052] In this disclosure, if articles are added through translation, such as a, an, and the in English, this disclosure may include the fact that the noun following these articles is plural.

[0053] In this disclosure, the term "A and B are different" may mean "A and B are different from each other." The term may also mean "A and B are each different from C." Terms such as "separate" and "combine" may be interpreted similarly to "different."

[0054] 10: Information processing device, 11: Acquisition unit, 12: Storage unit, 13: Generation unit, 14: Calculation unit, 15: Output unit, 1001: Processor, 1002: Memory, 1003: Storage, 1004: Communication device, 1005: Input device, 1006: Output device.

Claims

1. An information processing device comprising: an acquisition unit that acquires new campaign information including information on stores that will implement a new campaign and the type of business to which each store belongs; a hierarchical Bayesian model that includes stores that have implemented multiple campaigns in the past and the types of business to which each store belongs at different hierarchical levels; a variable that represents the effect according to the timing of when the campaign was implemented; a calculation unit that uses the acquired new campaign information to calculate effect information regarding the effect of the new campaign; and an output unit that outputs the calculated effect information.

2. In the hierarchical Bayesian model, y i is used as the estimated value of the causal effect of each campaign, θ i is the mean, and σ i 2 is the variance. Then, y i ~N(θ i , σ i 2 ). Let z ai be the j-dimensional store dummy variable, z bi be the i-dimensional business format dummy variable, x i be other explanatory variables, α be the intercept, θ a be the store effect, θ b be the business format effect, β be the coefficient of x i , τ a be the variance of θ a , τ b be the variance of θ b , z ci be the time dummy variable, and when a Gaussian process is set as the prior distribution for the effect corresponding to the period when the campaign is conducted for θ c , θ i = α + θ a T z ai + θ b T z bi + β T x i + θ c T z ci (Note that θ a T , z ai , θ b T , z bi , β T , x i , θ c T , z ci are all vectors) θ aj ~(0, τ a 2 ), j = 1, …, J θ bk ~(0, τ b 2 ), k = 1, …, K θ c ~N(0 N , C(σ 2 p , l p The information processing apparatus according to claim 1, characterized in that it is:

3. The store effect θ a and the aforementioned business type effect θ b The information processing apparatus according to claim 2, characterized in that the prior distribution for is a horseshoe prior distribution.

4. The information processing device according to claim 1, further comprising a generation unit that generates the hierarchical Bayesian model using information about the stores that have implemented multiple campaigns in the past and the types of businesses to which those stores belong, and information about the effectiveness of each of those campaigns.

5. An information processing method comprising: an acquisition step of acquiring new campaign information, including information on stores that will implement a new campaign and the type of business to which each store belongs; a calculation step of calculating effect information regarding the effect of the new campaign using a hierarchical Bayesian model that includes stores that have implemented multiple campaigns in the past and the types of business to which each store belongs at different hierarchical levels; a variable that represents the effect according to the timing of when the campaign was implemented; and an output step of outputting the calculated effect information.

Images