Purchase quantity estimation device and purchase quantity estimation program for regularly purchased products

Abstract

To optimally estimate a purchase quantity of regularly purchased products compared to the conventional technology, and, furthermore, accurately calculate the survival rate of customers purchasing regularly purchased products, taking into account the characteristics of regularly purchased products for which the explicit intent of enrollment and withdrawal cannot be confirmed to optimally estimate the purchase quantity of regularly purchased products compared to the conventional technology.SOLUTION: A purchase quantity estimation device for regularly purchased products acquires the survival rate of customers who continue purchasing the regularly purchased products, uses a predictive model that estimates the initial purchase quantity of the regularly purchased products based on the placement amount for each advertising medium where advertisements are placed for the regularly purchased products to acquire an initial purchase quantity prediction value of the regularly purchased products, and predicts the regular purchase quantity of the regularly purchased products using the initial purchase quantity prediction value and the survival rate.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to an apparatus for estimating the purchase quantity of regularly purchased goods and a program for estimating the purchase quantity.

Background Art

[0002] The survival rate of customers who purchase regularly purchased goods is important information when estimating the purchase quantity of regularly purchased goods.

[0003] However, regularly purchased goods have the characteristic that it is difficult to calculate the survival rate because it is impossible to determine "survival" or "death" by clearly confirming the intention of membership or cancellation, and it is difficult to calculate the survival rate.

[0004] That is, customers of regularly purchased goods often make choices such as stopping purchases temporarily or voluntarily ("suspending membership") (and then resuming purchases), or not purchasing the goods only for a specific time (purchasing the goods at other times). For this reason, months without customer purchases occur occasionally, so it is difficult to calculate the survival rate by regarding temporary or voluntary "withdrawal / cancellation" as an event occurrence.

[0005] Patent Document 1 (Japanese Unexamined Patent Application Publication No. 2020-190912) discloses that the cumulative number of media that an advertisement for a product or service has affected in each phase of the marketing funnel is obtained for each specific user, and based on the cumulative number and the purchase amount per specific user, the contribution degree of each media related to the purchase of the product or service is calculated, and based on the contribution degree of each media, the advertising appeal expression, and / or an estimation model generated based on the reason for continuous purchase of the product or service by a specific user, advertising planning for the product or service (such as the amount of advertisement submissions to each media) is determined, thereby forming, maintaining, and expanding regular purchasers who are the royal customer layer of mail-order and subscription businesses.

[0006] Patent Document 2 (Japanese Patent Application Laid-Open No. 2021-105838) stores a learning model in which the relationship between the behavior history of each of a plurality of users who have used a service in the past and the usage result of the service included in the behavior history is learned, obtains the behavior history of a user who is using the service, and based on the behavior history of the user who is using the service and the learning model, predicts the usage result of the user who is using the service.

[0007] Patent Document 3 (Japanese Patent Application Laid-Open No. 2015-64782) stores weights for each of the first medium on which an advertisement first viewed by a user is posted, the final medium on which an advertisement viewed immediately before conversion is posted, and the intermediate medium on which an advertisement viewed after the user first views an advertisement and before viewing the advertisement immediately before conversion is posted, identifies the first medium, the intermediate medium, and the final medium based on an advertisement log, and calculates a click point and a view point according to the number of clicks and the weight in each of the first medium, the intermediate medium, and the final medium.

Prior Art Documents

Patent Documents

[0008]

Patent Document 1

Patent Document 2

Patent Document 3

Summary of the Invention

Problems to be Solved by the Invention

[0009] An object of the present invention is to be able to optimally estimate the purchase quantity of regularly purchased products as compared with the prior art.

[0010] Furthermore, in consideration of the characteristics of regular purchase products for which it is impossible to confirm a clear intention to join or cancel, the present invention aims to accurately calculate the survival rate of customers who purchase regular purchase products, and thus be able to optimally estimate the purchase volume of regular purchase products as compared with the prior art.

Means for Solving the Problems

[0011] A first aspect is a purchase volume estimation device for regular purchase products, comprising: a survival rate acquisition unit that acquires the survival rate of customers who continue to purchase regular purchase products; a first purchase volume prediction value acquisition unit that acquires a predicted value of the first purchase volume of regular purchase products using a prediction model that estimates the first purchase volume of regular purchase products from the publication volume for each advertisement publication medium for regular purchase products; and an estimation unit that predicts the regular purchase volume of regular purchase products using the predicted value of the first purchase volume acquired by the first purchase volume prediction value acquisition unit and the survival rate.

[0012] A second aspect is a purchase volume estimation device for regular purchase products according to the first aspect, wherein the estimation unit predicts the first purchase volume of regular purchase products and predicts the regular purchase volume of regular purchase products such that the total purchase volume obtained by summing the first purchase volume and the regular purchase volume of regular purchase products is maximized.

[0013] A third aspect is an order receiving volume estimation device for regular purchase products according to the first aspect, wherein the estimation unit predicts the first purchase volume of regular purchase products and predicts the regular purchase volume of regular purchase products such that the total purchase volume obtained by summing the first purchase volume and the regular purchase volume of regular purchase products becomes the target purchase volume.

[0014] A fourth aspect is a purchase volume estimation device for regular purchase products according to the first aspect, comprising: an actual value acquisition unit that acquires the actual value of the first purchase volume of regular purchase products and the actual value of the publication volume for each advertisement publication medium for regular purchase products; and a prediction model generation unit that generates a prediction model for estimating the first purchase volume of regular purchase products from the publication volume for each advertisement publication medium using the actual value of the first purchase volume and the actual value of the publication volume.

[0015] A fifth aspect is, in the first aspect, that the survival rate acquisition unit includes a non-effective month number acquisition unit that acquires the non-effective month number for determining that a customer who has made a first purchase of a regularly purchased product has stopped making regular purchases, and for each customer who has made a first purchase of a regularly purchased product, compares the non-effective month number with the number of months elapsed since the final purchase date of the regularly purchased product to determine whether the customer who has made a first purchase of the regularly purchased product has stopped making regular purchases, a regular purchase stop determination unit, for each customer who has made a first purchase of a regularly purchased product, calculates the survival period from the first purchase date of the regularly purchased product to the final purchase date of the regularly purchased product, a survival period calculation unit, and uses the survival period of each customer who has made a first purchase of the regularly purchased product to calculate the survival rate at which the customer continues to purchase the regularly purchased product, a survival rate calculation unit, and is a regularly purchased product purchase quantity estimation device including these components.

[0016] A sixth aspect is a regularly purchased product purchase quantity estimation program for causing a computer to execute the following processing in order to estimate the purchase quantity of a regularly purchased product. The program includes a non-effective month number acquisition process for acquiring the non-effective month number for determining that a customer who has made a first purchase of a regularly purchased product has stopped making regular purchases, a regular purchase stop determination process for comparing the non-effective month number with the number of months elapsed since the final purchase date of the regularly purchased product for each customer who has made a first purchase of the regularly purchased product to determine whether the customer who has made a first purchase of the regularly purchased product has stopped making regular purchases, a survival period calculation process for calculating the survival period from the first purchase date of the regularly purchased product to the final purchase date of the regularly purchased product for each customer who has made a first purchase of the regularly purchased product, a survival rate calculation process for calculating the survival rate at which the customer continues to purchase the regularly purchased product using the survival period of each customer who has made a first purchase of the regularly purchased product, a first purchase quantity predicted value acquisition process for acquiring the predicted value of the first purchase quantity of the regularly purchased product, and an estimation process for estimating the predicted value of the regular purchase quantity of the regularly purchased product or / and the predicted value of the total purchase quantity obtained by summing the first purchase quantity and the regular purchase quantity of the regularly purchased product using the survival rate and the predicted value of the first purchase quantity.

[0017] A seventh aspect is a purchase quantity estimation device for regularly purchased goods that estimates the purchase quantity of regularly purchased goods, comprising: a non-effective month number acquisition unit that acquires the number of non-effective months for determining that a customer who made a first purchase of a regularly purchased good has stopped making regular purchases; a regular purchase stop determination unit that compares the number of non-effective months with the number of months elapsed since the final purchase date of the regularly purchased good for each customer who made a first purchase of the regularly purchased good, and determines that the customer who made a first purchase of the regularly purchased good has stopped making regular purchases; a survival period calculation unit that calculates the survival period from the first purchase date of the regularly purchased good to the final purchase date of the regularly purchased good for each customer who made a first purchase of the regularly purchased good; a survival rate calculation unit that calculates the survival rate at which the customer continues to purchase the regularly purchased good using the survival period of each customer who made a first purchase of the regularly purchased good; a first purchase quantity prediction value acquisition unit that acquires a predicted value of the first purchase quantity of the regularly purchased good; and an estimation unit that estimates a predicted value of the regular purchase quantity of the regularly purchased good and / or a predicted value of the total purchase quantity obtained by summing the first purchase quantity and the regular purchase quantity of the regularly purchased good using the survival rate and the predicted value of the first purchase quantity.

[0018] An eighth aspect is the purchase quantity estimation device for regularly purchased goods according to the seventh aspect, wherein the non-effective month number acquisition unit includes a non-effective month number reception unit that receives the number of non-effective months on a display screen of a display device of a computer.

[0019] A ninth aspect is the purchase quantity estimation device for regularly purchased goods according to the seventh aspect, wherein the estimation unit predicts the first purchase quantity of the regularly purchased good using a prediction model that estimates the first purchase quantity of the regularly purchased good from the quantity of each advertisement published for the regularly purchased good, and predicts the regular purchase quantity of the regularly purchased good using the predicted first purchase quantity and the survival rate.

Advantages of the Invention

[0020] According to the first aspect to the ninth aspect, compared with the prior art, the purchase quantity of regularly purchased products can be optimally estimated. Further, according to the first aspect to the ninth aspect, considering the characteristics of regularly purchased products for which it is not possible to confirm a clear intention to join or leave, the survival rate of customers who purchase regularly purchased products can be accurately calculated, and thus, compared with the prior art, the purchase quantity of regularly purchased products can be optimally estimated.

Brief Description of the Drawings

[0021]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Embodiment for Carrying Out the Invention

[0022] Hereinafter, with reference to the drawings, embodiments of a purchase quantity estimation device and a purchase quantity estimation program for regularly purchased goods according to the present invention will be described.

[0023] (Functional Configuration of Purchase Quantity Estimation Device for Regularly Purchased Goods)

[0024] Figure 1 is a block diagram showing the functional configuration of a purchase quantity estimation device 100 for regularly purchased goods according to the embodiment.

[0025] The purchase quantity estimation device 100 for regularly purchased goods according to the embodiment generally includes a survival rate acquisition unit 160, a first purchase quantity predicted value acquisition unit 190, and an estimation unit 200.

[0026] The survival rate acquisition unit 160 includes a non-effective month number acquisition unit 120, a regular purchase stop determination unit 130, a survival period calculation unit 140, and a survival rate calculation unit 150. The non-effective month number acquisition unit 120 further includes a non-effective month number reception unit 110.

[0027] The first purchase quantity predicted value acquisition unit 190 includes an actual value acquisition unit 170 and a prediction model generation unit 180.

[0028] The functions in Figure 1 can, as shown in Figure 2, be realized by a combination of one or more personal computer terminals 60, a server 70, and a network 80 that communicably connects each personal computer terminal 60 and the server 70. It can also be realized by a single personal computer terminal 60.

[0029] (Hardware Configuration)

[0030] Figure 3 is a hardware configuration diagram for realizing the functional configuration of the embodiments shown in FIGS. 1 and 2. The hardware configuration of the personal computer terminal 60 or the server 70 is illustrated.

[0031] As shown in FIG. 3, the personal computer terminal 60 or the server 70 includes a CPU (Central Processing Unit) 61, a ROM (Read OnlY MemorY) 62, a RAM (Random Access MemorY) 63, a storage 64, an input device 66, a display device 67, a communication I / F (interface) 68, and an external storage device 69, and are communicably connected to each other via a system bus 65.

[0032] The CPU 61 is a central arithmetic processing unit that executes various programs and controls each device connected to the system bus 65. That is, the CPU 61 reads a program from the ROM 62 or the storage 64 and executes the program using the RAM 63 as a work area. The CPU 61 controls each device connected to the system bus 65 and performs various arithmetic processes according to the programs recorded in the ROM 62 or the storage 64. The ROM 62 or the storage 64 holds a BIOS (Basic Input / Output SYstem), an OS (Operating SYstem), which are control programs executed by the CPU 61, various programs that can be read and executed by a computer for realizing the present embodiment, and various necessary data.

[0033] The ROM 62 stores various control programs and various data. The RAM 63 functions as a main memory, a work area, etc. of the CPU 61 and temporarily stores a program or data as a work area. The storage 64 is composed of an HDD (Hard Disk Drive) or an SSD (Solid State Drive) and stores various programs and various data including the BIOS and the OS.

[0034] The input device 66 includes a pointing device such as a mouse and reading devices such as a keyboard and a scanner, and is used to perform various inputs.

[0035] The display device 67 is, for example, a liquid crystal display and displays various information. The display device 67 may adopt a touch panel method and function as the input device 66.

[0036] The communication I / F (interface) 68 is an interface for communicating with devices such as other servers 70 and terminals 60. For example, standards such as Ethernet (registered trademark), FDDI, and Wi-Fi (registered trademark) are used. The communication I / F (interface) 68 is connected to the network 80 and controls the transmission and reception of data.

[0037] The external storage device 69 is composed of various memory cards such as USB memories and externally attachable storage media such as HDDs and SSDs.

[0038] (Purchase Quantity Estimation Program for Regularly Purchased Items)

[0039] Figure 4 shows, in a flowchart, the processing procedure of the purchase quantity estimation program PB100 for regularly purchased items to realize the functions of the purchase quantity estimation device 100 for regularly purchased items.

[0040] The purchase quantity estimation program PB100 for regularly purchased items causes a computer to execute a non-effective month number acquisition process S120, a regular purchase stop determination process S130, a lifetime calculation process S140, a survival rate calculation process S150, a first purchase quantity prediction value acquisition process S190, and an estimation process S200.

[0041] The invalid month count acquisition process S120 corresponds to the process performed by the invalid month count acquisition unit 120. The regular purchase stop determination process S130 corresponds to the process performed by the regular purchase stop determination unit 130. The survival period calculation process S140 corresponds to the process performed by the survival period calculation unit 140. The survival rate calculation process S150 corresponds to the process performed by the survival rate calculation unit 150. The initial purchase quantity predicted value acquisition process S190 corresponds to the process performed by the initial purchase quantity predicted value acquisition unit 190. The estimation process S200 corresponds to the process performed by the estimation unit 200.

[0042] The purchase quantity estimation program PB100 for regularly purchased goods is stored in the ROM 62, the storage 64, or the external storage device 69. The personal computer terminal 60 or the server 70 reads out the purchase quantity estimation program PB100 for regularly purchased goods from the ROM 62, the storage 64, or the external storage device 69 and executes each of the above processes.

[0043] Note that it can be arbitrarily determined according to the system configuration whether each of the above processes S120 to S200 is performed by the personal computer terminal 60 or the server 70. For example, the purchase quantity estimation program PB100 for regularly purchased goods may be installed in the personal computer terminal 60, and the purchase quantity estimation program PB100 for regularly purchased goods may be executed by the personal computer terminal 60 alone. Also, for example, by accessing the server 70 from the personal computer terminal 60 via the network 80, the processes of the purchase quantity estimation program PB100 for regularly purchased goods may be executed by both the personal computer terminal 60 and the server 70.

[0044] The purchase quantity estimation device 100 for regularly purchased goods according to the embodiment executes the purchase quantity estimation program PB100 for regularly purchased goods according to the processing procedure shown in FIG. 4.

[0045] (Invalid month count acquisition process S120)

[0046] The invalid month count acquisition unit 120 executes a process of acquiring the invalid month count NI.

[0047] The invalid month number reception unit 110 receives the invalid month number NI on the display screen of the display device 67 of the personal computer terminal (hereinafter referred to as the terminal) 60. The invalid month number NI is the number of months that serves as a threshold for determining that a customer who has made a first purchase of a regularly purchased product has stopped the regular purchase. When a customer has not purchased a regularly purchased product even after the invalid month number NI has elapsed since the previous purchase, it is regarded as having stopped the regular purchase.

[0048] FIG. 5 illustrates a display screen 310 as an invalid month number reception screen of the display device 67 of the terminal 60. The invalid month number reception screen 310 is configured as an interface screen having the function of the invalid month number reception unit 110.

[0049] On the invalid month number reception screen 310, a period setting unit 311 and a target product classification setting unit 312 are arranged.

[0050] The period setting unit 311 is composed of, for example, a text box, and the start date Ys year Ms month Ds day (for example, June 1, 2019), the end date Ye year Me month De day (for example, August 31, 2022) of the analysis target period, and the invalid month number NI (for example, 3.5 months) can be text-input.

[0051] The target product classification setting unit 312 is composed of, for example, a text box, and the classification of the regularly purchased product to be analyzed (for example, lactic acid bacteria supplement) can be text-input.

[0052] The data including the invalid month number NI received on the invalid month number reception screen 310 is stored in the ROM 62 or the storage 64 of the server 70 or the external storage device 69.

[0053] (Regular purchase stop determination process S130)

[0054] The regular purchase stop determination unit 130 compares the invalid month number NI and the number of elapsed months NC from the final purchase date NE of the regularly purchased product for each customer who has made a first purchase of the regularly purchased product, and determines that the customer who has made a first purchase of the regularly purchased product has stopped the regular purchase.

[0055] FIG. 6 illustrates a display screen 51 as a sales data screen 320 of the display device 67 of the terminal 60.

[0056] A sales data display section 321 is arranged on the sales data screen 320.

[0057] The sales data display section 321 displays sales data 321D including a number (hereinafter referred to as customer number) ID for identifying a purchased customer in association with the purchase date (IDs of individual customers are shown as ID1, ID2, ID3, ···), the purchase date, the number of purchases, and the purchased products. The sales data 321D is stored, for example, in the ROM 62 or the storage 64 or the external storage device 69 of the server 70. The sales data 321D stored in the ROM 62 or the storage 64 or the external storage device 69 of the server 70 is read out and displayed on the sales data screen 320 of the display device 67 of the terminal 70.

[0058] The regular purchase stop determination section 130 determines, for each customer, whether or not the regular purchase has been stopped based on the sales data 321D.

[0059] For example, it is shown on the sales data display section 321 that customer ID1 initially purchased 3 regular purchase products on the first purchase date NS (June 1, 2019), purchased 2 regular purchase products on July 2, 2019, and purchased 1 regular purchase product on August 1, 2019, but as of November 16, 2019, no regular purchase products have been purchased after August 2, 2019. Therefore, for customer ID1, the invalid months NI (3.5 months) and the elapsed months NC (3.5 months) from the final purchase date NE (August 1, 2019) of the regular purchase product are compared, and since the elapsed months NC (3.5 months) are equal to or more than the invalid months NI (3.5 months), it is determined that customer ID1 who initially purchased the regular purchase product stopped the regular purchase with the final purchase date NE (August 1, 2019). Whether or not the regular purchase of the regular purchase product has been stopped for other customer IDs ID2, ID3, ··· is determined in the same way.

[0060] By performing this determination, purchase date data BD including the first purchase date NS and the last purchase date NE for each customer number ID is obtained.

[0061] The obtained purchase date data BD is stored in the ROM 62 or the storage 64 or the external storage device 69 of the server 70.

[0062] (Lifetime calculation process S140)

[0063] The lifetime calculation unit 140 calculates the lifetime LT for each customer who made the first purchase of a regularly purchased product.

[0064] FIG. 7 illustrates a display screen 51 as a customer-specific data screen 330 of the display device 67 of the terminal 60.

[0065] A customer-specific data display unit 331 is arranged on the customer-specific information screen 330.

[0066] On the customer-specific data display unit 331, customer-specific data 331D including the first purchase date NS, the last purchase date NE, and the lifetime LT is displayed for each customer number ID (each individual customer is indicated by ID1, ID2, ID3,...).

[0067] The purchase date data BD stored in the ROM 62 or the storage 64 or the external storage device 69 of the server 70 is read out, and the first purchase date NS and the last purchase date NE are displayed on the customer-specific data screen 330 of the display device 67 of the terminal 70. The lifetime LT is calculated by the lifetime calculation unit 140 as the period from the first purchase date NS to the last purchase date NE. The calculated lifetime LT is displayed on the customer-specific data screen 330 of the display device 67 of the terminal 70. For example, for customer ID1, a lifetime LT (2.0 months) is calculated according to the first purchase date NS (June 1, 2019) and the last purchase date NE (August 1, 2019), and is displayed on the customer-specific data screen 330.

[0068] The customer-specific data 331D includes the number of purchases, the total purchase amount, the total number of purchases, the average purchase interval, the purchase quantity by elapsed months, and the purchase amount by elapsed months.

[0069] The customer-specific data 331D is stored in the ROM 62 or the storage 64 or the external storage device 69 of the server 70.

[0070] (Survival rate calculation process S150)

[0071] The survival rate calculation unit 150 calculates the survival rate RS using the survival period LT calculated by the survival period calculation unit 140. Here, the survival rate RS is the remaining rate at which the customer continues to purchase the regularly purchased product.

[0072] FIG. 8 illustrates a display screen 51 as a survival rate display screen 340 of the display device 67 of the terminal 60.

[0073] A survival rate display unit 341 is arranged on the survival rate display screen 340.

[0074] Hereinafter, the process of calculating the survival rate RS will be described with reference to FIG. 8.

[0075] The customer-specific data 331D stored in the ROM 62 or the storage 64 or the external storage device 69 of the server 70 is read out.

[0076] The customer-specific data 331D includes data indicating the first purchase date NS, the last purchase date NE, and the survival period LT for each customer ID.

[0077] Based on the data of the first purchase date NS, the last purchase date NE, and the survival period LT for each customer number ID, the number of customers who continue to purchase the regularly purchased product can be calculated for each number of months YM. Based on the number of customers who continue to purchase the regularly purchased product for each number of months YM, the ratio RD of the number of customers who continue to make regular purchases can be calculated in association with the elapsed months MG.

[0078] For example, assuming that the number of months YM since the first purchase by a customer is October 2019, the ratio of the number of customers who continue to purchase regularly (cumulative survival rate) RD1 is calculated to be 45% six months later, which is the number of elapsed months MG.

[0079] In this way, the ratio of the number of customers (cumulative survival rate) RD is calculated in association with the number of months YM (vertical axis of the survival rate display section 341) and the number of elapsed months MG (horizontal axis of the survival rate display section 341). Note that the ratio of the number of customers (cumulative survival rate) RD0 corresponding to the number of elapsed months MG being 0 months (MG0) is 100%, indicating the ratio of the number of customers who have only made the first purchase of the regularly purchased product (for example, the ratio of the number of customers who have only made the first purchase of the regularly purchased product in October 2019 is 18.7%).

[0080] The cumulative survival rate may be calculated using the known Cutler-Ederer method. In this case, it is obtained as follows by a calculation method considering the characteristics of the regularly purchased product. Hereinafter, survival is used in the sense of continued purchase, and death is used in the sense of termination of purchase.

[0081] 1) Since changes are recognized in the survival curve in a time series, the cumulative survival rate is calculated for each first purchase interval.

[0082] For example, for the number of months YM of "October 2019", the survival rate for the first month is obtained as the ratio of the initial survival number (the survival number on October 1, 2019) and the effective survival number obtained by subtracting the number of deaths during the period (the number of deaths by October 31, 2019) from the initial survival number (the survival number on October 1, 2019). Thereafter, the survival rate is calculated for each of the second month period, the third month period, and so on. The calculated survival rates are sequentially multiplied to calculate the cumulative survival rate. The cumulative survival rate for the second month period is obtained as the product of the survival rate for the first month period (0.87) and the survival rate for the second month period (0.69). Similarly, for the periods after the third month, the same applies. In this way, the cumulative survival rates (0.87, 0.69, 0.63, 0.54, 0.51, 0.45,...) for each number of elapsed months MG for the number of months YM (October 2019) are calculated.

[0083] Similarly, for other monthly counts YM (November 2019), monthly counts YM (December 2019), etc., the cumulative survival rate for each elapsed month count MG is calculated for each monthly count YM.

[0084] 2) If no purchase occurs even after the specified period, i.e., the invalid month count NI has elapsed, it is determined that the survival has ended (death). This is due to the characteristics of the recurring purchase product offering service where a clear indication of the intention to withdraw cannot be confirmed.

[0085] 3) Even if a purchase resumes after being determined that the survival has ended, it is regarded as having died and not counted as a survival number.

[0086] 4) Since the survival numbers for only the first purchase vary significantly depending on the measures, the survival for only the first purchase is excluded.

[0087] For example, for the monthly count YM of "October 2019", the ratio for only the first purchase was 18.7%

[0088] However, the survival numbers for this ratio are excluded and the cumulative survival rate is calculated.

[0089] 5) Using the reduction rate of the survival rate and its predicted value, interpolation (prediction) of the cumulative survival rate for each first purchase time is performed.

[0090] For example, for the monthly count YM of "August 2022", up to the 3 - month period, the cumulative survival rate can be calculated by successively multiplying the survival rates. However, for the period after the 4 - month period, since the survival rate to be multiplied has not been calculated, the cumulative survival rate cannot be calculated. In such a case, the predicted survival rate is calculated from the reduction rate of the survival rate calculated for other monthly counts YM, and the cumulative survival rate for the period after the 4 - month period is interpolated.

[0091] The cumulative survival rate may also be calculated using the known Kaplan-Meier method. For example, every time death (purchase completion) occurs, the survival rate may be sequentially calculated from the ratio of the initial survival number to the current effective survival number, and the cumulative survival rate may be calculated by sequentially multiplying the survival rates. Also in this case, the cumulative survival rate is calculated by a calculation method that takes into account the characteristics of the regularly purchased product in the same manner as in the above 1) to 5).

[0092] The survival rate RS is calculated as the average value RDa of the cumulative survival rates RD (RD1, RD2, RD3, ··· RDN) at the elapsed months MG in the same month (for example, after 6 months) calculated for each number of months YM.

[0093] The survival rate RS (for example, the survival rate corresponding to the elapsed months MG (for example, after 6 months)) is calculated as the average value RDa (= 45%) obtained by averaging the cumulative survival rates RD1 (= 45%), RD2 (= 54%), RD3 (= 50%), ··· RDN (= 43%) at the elapsed months MG in the same month (for example, after 6 months) for each of the number of months YM for N months, as shown in the following formula (10).

[0094] RS = RDa = (RD1 + RD2 + RD3 + ··· + RDN) / N

[0095] ···(10)

[0096] FIG. 9 illustrates a graph of the survival rate RS (vertical axis) according to the elapsed months MG (horizontal axis).

[0097] The data RSD indicating the survival rate RS is stored in the ROM 62, the storage 64, or the external storage device 69 of the server 70.

[0098] (Initial purchase quantity prediction value acquisition process S190)

[0099] The initial purchase quantity prediction value acquisition unit 190 acquires the initial purchase quantity prediction value Yd of the regularly purchased product.

[0100] The initial purchase quantity prediction value acquisition unit 190 obtains a predicted value Yd of the initial purchase quantity of the regularly purchased product by using a prediction model 400 that estimates the initial purchase quantity Y of the regularly purchased product from the publication quantities X1, X2, X3, X4, and X5 for each of the advertisement publication media 1, 2, 3, 4, and 5 for which advertisements are published for the regularly purchased product.

[0101] The actual value acquisition unit 170 obtains the actual value Z of the initial purchase quantity of the regularly purchased product and the actual values W1, W2, W3, W4, and W5 of the publication quantities for each of the advertisement publication media 1, 2, 3, 4, and 5 for which advertisements are published for the regularly purchased product.

[0102] FIG. 10 illustrates a display screen 51 as an actual value screen 350 of the display device 67 of the terminal 60. The actual value screen 350 is configured as an interface screen having the function of the actual value acquisition unit 170. For example, when the user performs an operation on the actual value screen 350 to obtain the advertisement publication actual values W1, W2, W3, W4, and W5 and the initial purchase quantity actual value Z, the obtained advertisement publication actual values W1, W2, W3, W4, and W5 and the initial purchase quantity actual value Z are displayed on the actual value display unit 351 of the actual value screen 350. For example, the advertisement publication actual values W1, W2, W3, W4, and W5 for the analysis target period and the initial purchase quantity actual value Z for the said period are read from the database (ROM 62 or storage 64 or external storage device 69) of the server 70 and displayed on the actual value display unit 351 of the actual value screen 350. The advertisement publication actual values W1, W2, W3, W4, and W5 and the initial purchase quantity actual value Z are used for the calculation for the generation of the prediction model 400. Note that the "···" in the cells in the table of the actual value display unit 351 in FIG. 10 indicates specific numerical values.

[0103] On the other hand, the initial purchase quantity actual value Z can be, for example, the purchase quantity by elapsed months. Also, the initial purchase quantity actual value Z may be the purchase amount by elapsed months. The purchase quantity by elapsed months and the purchase amount by elapsed months can be obtained from the customer-specific data 331D.

[0104] Define the variables of the publication volumes of each advertising publication medium 1, 2, 3, 4, and 5 as X1, X2, X3, X4, and X5 respectively. The publication volume is used in the sense of including the number of publications and the publication amount. Also, define the variable of the first-time purchase volume of regularly purchased products as Y.

[0105] The advertising publication actual values W1, W2, W3, W4, and W5 respectively indicate the actual values of the publication volumes X1, X2, X3, X4, and X5 of the advertising publication media 1, 2, 3, 4, and 5. The first-time purchase volume actual value Z indicates the actual value of the first-time purchase volume Y.

[0106] Each of the advertising publication media 1, 2, 3, 4, and 5 is, for example, an advertisement in a local newspaper, an advertisement in a national newspaper, an advertisement in over-the-air television broadcasting, an advertisement in radio broadcasting, or a leaflet insert. These are just examples, and as advertising publication media, any media capable of advertising publication can be arbitrarily selected and combined. For example, when the advertising publication media is a newspaper advertisement, advertisements in different newspapers or newspapers in different regions can be selected and combined, or different publication forms such as the publication size, black-and-white advertisement, and color advertisement can be selected and combined. Also, for example, when the advertising publication media is an advertisement in television broadcasting, advertisements in different broadcasting forms such as over-the-air television broadcasting, satellite broadcasting, and cable broadcasting can be selected and combined, or specific or arbitrary TV commercials or TV programs can be selected and combined. Also, for example, when the advertising publication media is an advertisement in radio broadcasting, for example, arbitrary radio commercials or radio programs can be selected and combined. Also, for example, when the advertising publication media is an advertisement on the Internet, for example, specific or arbitrary digital advertisements, websites, SNSs, and application software installed on information terminals such as smartphones and personal computers can be selected and combined. Also, for example, when the advertising publication media is an outdoor advertisement, advertisements inside vehicles in various moving media such as billboards, digital signage, trains, buses, and automobiles can be selected and combined. Any one of these newspaper advertisements, advertisements in television broadcasting, advertisements in radio broadcasting, advertisements on the Internet, and outdoor advertisements can be selected, or two or more can be selected and combined.

[0107] The prediction model generation unit 180 generates a prediction model 400 that estimates the initial purchase quantity Y of a regularly purchased product from the publication quantities X1, X2, X3, X4, X5 for each of the advertisement publication media 1, 2, 3, 4, 5, using the actual initial purchase quantity value Z and the actual publication quantity values W1, W2, W3, W4, W5.

[0108] Hereinafter, the prediction model 400 will be described.

[0109] For the prediction model 400, analysis methods such as linear regression analysis like multiple regression analysis can be applied. Also, for the prediction model 400, analysis methods of non - linear regression analysis such as logistic regression analysis and polynomial regression analysis can be applied.

[0110] The prediction model 400 can be expressed, for example, by the following formula (1) or the following formula (2).

[0111] Y = Γ / {1 + exp(β0 + β1·X1 + β2·X2 + β3·X3 + β4·X4 + β5·X5)}

[0112] ···(1)

[0113] Y = β0 + β1·X1 + β2·X2 + β3·X3 + β4·X4 + β5·X5

[0114] ···(2)

[0115] Here, the above formula (1) represents the formula of a prediction model that is a logistic regression analysis model. The above formula (2) represents the formula of a prediction model to which multiple regression analysis is applied. The regression coefficients β0, β1, β2, β3, β4, and β5 are obtained by applying the least squares method to, for example, the data of the advertising release performance values W1, W2, W3, W4, W5 and the first purchase quantity performance value Z. For example, regression coefficients β0, β1, β2, β3, β4, and β5 that minimize the average of the squared errors are obtained using the data of the advertising release performance values W1, W2, W3, W4, W5 and the first purchase quantity performance value Z. Note that the logistic regression analysis model can be applied with multiple regression analysis by performing variable transformation. The upper limit value of the logistic regression analysis model is defined by the constant Γ (referred to as a hyperparameter constant) on the right side of the above formula (1). It is necessary to determine the hyperparameter constant Γ before performing the multiple regression analysis for estimating the regression coefficients β0 to β5.

[0116] The first purchase quantity predicted value acquisition unit 190 acquires the first purchase quantity predicted value Yd by substituting the release quantities X1 to X5 of the prediction model 400 generated by the prediction model generation unit 180 with the release quantities corresponding to the release budget.

[0117] Also, a target value Yd of the first purchase quantity predicted value may be determined, the target value Yd may be substituted into the first purchase quantity Y that is the target variable of the prediction model 400, and the optimal release quantities Xd1, Xd2, Xd3, Xd4, Xd5 that minimize the sum of the explanatory variables X1, X2, X3, X4, X5 may be calculated.

[0118] Also, the optimal combination of the release quantities of the explanatory variables X1, X2, X3, 4, X5 of the prediction model 400 may be calculated as the optimal release quantities Xd1, Xd2, Xd3, Xd4, Xd5 so that the first purchase quantity predicted value Yd that is the target variable of the prediction model 400 is maximized.

[0119] (Estimation process S200)

[0120] The estimation unit 200 estimates the predicted regular purchase quantity Yds of regularly purchased products or / and the predicted total purchase quantity YdT obtained by summing the initial purchase quantity Yd and the regular purchase quantity Yds of regularly purchased products, using the survival rate RS and the predicted value Yd of the initial purchase quantity.

[0121] FIG. 11 illustrates a display screen 51 as a regular purchase quantity estimation screen 360 of the display device 67 of the terminal 60. The regular purchase quantity estimation screen 360 is configured as an interface screen having the functions of the estimation unit 200.

[0122] On the regular purchase quantity estimation screen 360, a regular purchase quantity estimated value display unit 361 is arranged.

[0123] In the regular purchase quantity estimated value display unit 361, the predicted value Yd of the initial purchase quantity, the predicted value Yds of the regular purchase quantity, and the predicted value YdT of the total purchase quantity are displayed for each future number of months YM.

[0124] The predicted value Yd of the initial purchase quantity can be obtained using the prediction model 400. The survival rate RS can be obtained by reading the survival rate data RSD from the ROM 62 or the storage 64 or the external storage device 69 of the server 70.

[0125] The predicted value Yds of the regular purchase quantity is obtained by cumulatively adding the purchase quantity obtained by multiplying the predicted value Yd of the initial purchase quantity by the survival rate RS according to the number of months YM.

[0126] The total purchase predicted value YdT is obtained by adding the predicted value Yd of the initial purchase quantity and the predicted value Yds of the regular purchase quantity.

[0127] FIG. 12 is a graph showing the relationship between the number of months YM and the predicted value Yd of the initial purchase quantity when the horizontal axis is the number of months YM and the vertical axis is the predicted value Yd of the initial purchase quantity.

[0128] Hereinafter, n is used in the sense of the number of months (for example, September 2022).

[0129] Define the predicted value Yd of the initial purchase quantity at the number of months YM(n) as Yd(n).

[0130] Also, define the predicted value of the regular purchase quantity Yds in the number of months and years YM(n) as Yds(n).

[0131] Also, define the survival rate RS when the number of elapsed months is 1 month as RS(1).

[0132] Therefore, Yd(n - 1) is the predicted value Yd of the first purchase quantity in the number of months and years YM(n - 1) one month before the number of months and years YM(n) (for example, in August 2022). The same applies when the number of months n decreases by 1 each time.

[0133] Also, Yds(n - 1) is the predicted value Yds of the regular purchase quantity in the number of months and years YM(n - 1) one month before the number of months and years YM(n) (for example, in August 2022). The same applies when the number of months n decreases by 1 each time.

[0134] Also, RS(2) is the survival rate RS when the number of elapsed months MG is 1 month later. The same applies when the number of months n increases by 1 each time.

[0135] Therefore, the predicted value of the regular purchase quantity Yds(n) in the number of months and years YM(n) is expressed by the following formula (11).

[0136] Yds(n)=Yd(n - 1)×RS(1)+Yd(n - 2)×RS(2)

[0137] +Yd(n - 3)×RS(3)+Yd(n - 4)×RS(4)

[0138] +Yd(n - 5)×RS(5)+Yd(n - 6)×RS(6)

[0139] +···

[0140] ···(11)

[0141] Define the predicted value of the total purchase quantity YdT in the number of months and years YM(n) as YdT(n).

[0142] The predicted total purchase quantity YdT(n) for the number of months YM(n) is represented by the following equation (12).

[0143] YdT(n) = Yd(n) + Yds(n)

[0144] ···(12)

[0145] Using the above equations (11) and (12), the predicted regular purchase quantity Yds and the predicted total purchase quantity YdT for future months YM are calculated respectively.

[0146] For example, in the regular purchase quantity estimation value display section 361, it is displayed that the predicted first - time purchase quantity Yd for the future month number YM "September 2022" is "425", the predicted regular purchase quantity Yds is "3596", and the predicted total purchase quantity YdT is "4021".

[0147] Figure 13 illustrates a graph with the number of months YM on the horizontal axis and the predicted first - time purchase quantity Yd, the predicted regular purchase quantity Yds, and the predicted total purchase quantity YdT on the vertical axis. The predicted first - time purchase quantity Yd is indicated by line L1, the predicted regular purchase quantity Yds is indicated by line L2, and the predicted total purchase quantity YdT is indicated by line L3. Note that lines L´1, L´2, and L´3 correspond to the actual first - time purchase quantity value, the actual regular purchase quantity value, and the actual total purchase quantity value respectively.

[0148] For example, the predicted first - time purchase quantity Yd (line L1) is calculated according to the amount of publication corresponding to the publication budget. When the predicted first - time purchase quantity Yd (line L1) is obtained according to the publication budget, the predicted regular purchase quantity Yds (line L2) and the predicted total purchase quantity YdT (line L3) can be obtained according to this predicted first - time purchase quantity Yd (line L1). In this case, the predicted first - time purchase quantity Yd (line L1) may be calculated so that the predicted first - time purchase quantity is maximized. Also in this case, when the maximum value of the predicted first - time purchase quantity Yd (line L1) is obtained, the predicted regular purchase quantity Yds (line L2) and the predicted total purchase quantity YdT (line L3) can be obtained according to this maximum value of the predicted first - time purchase quantity Yd (line L1).

[0149] Also, a target value Yd (line L1) of the predicted value of the initial purchase quantity may be calculated. Also in this case, when the predicted value of the initial purchase quantity Yd (line L1), which is the target value, is obtained, the predicted value of the regular purchase quantity Yds (line L2) and the predicted value of the total purchase quantity YdT (line L3) can be obtained according to the predicted value of the initial purchase quantity Yd (line L1), which is the target value.

[0150] Also, a target value YdT (line L3) of the predicted value of the total purchase quantity may be set. When the target value YdT (line L3) of the predicted value of the total purchase quantity is set, the predicted value of the initial purchase quantity Yd (line L1) and the predicted value of the regular purchase quantity Yds (line L2) can be obtained according to the target value YdT (line L3) of the predicted value of the total purchase quantity. When the predicted value of the initial purchase quantity Yd is obtained, the value Yd may be substituted into the initial purchase quantity Y, which is the target variable of the prediction model 400, and the optimal publication quantities Xd1, Xd2, Xd3, Xd4, Xd5 at which the sum of the explanatory variables, the publication quantities X1, X2, X3, X4, X5, becomes minimum may be calculated.

[0151] The "regular purchase product" in the present embodiment is defined as a concept including products and services that are regularly purchased. That is, the "regular purchase product" in the present embodiment includes not only articles that are regularly purchased but also services that are regularly purchased and provided. For example, the "regular purchase product" in the present embodiment includes a regular subscription service and a regular use service (so-called subscription type service) that can be suspended or temporarily stopped.

Explanation of Signs

[0152] 100 Purchase quantity estimation device for regular purchase products 110 Invalid month reception unit 120 Invalid month acquisition unit 130 Regular purchase stop determination unit 140 Lifetime calculation unit 150 Survival rate calculation unit 160 Survival rate acquisition unit 170 Actual value acquisition unit 180 Prediction model generation unit 190 Initial purchase quantity predicted value acquisition unit 200 Estimation unit

Claims

1. A survival rate acquisition unit that acquires the survival rate of customers who continue to purchase regularly purchased products; An initial purchase quantity prediction value acquisition unit that acquires a predicted value of the initial purchase quantity of regularly purchased products using a prediction model that estimates the initial purchase quantity of regularly purchased products from the publication volume for each advertising publication medium for which advertisements have been published for regularly purchased products; An estimation unit that predicts the regular purchase quantity of regularly purchased products using the predicted value of the initial purchase quantity acquired by the initial purchase quantity prediction value acquisition unit and the survival rate; A purchase quantity estimation device for regularly purchased products, comprising:

2. The estimation unit predicts the initial purchase quantity of regularly purchased products and predicts the regular purchase quantity of regularly purchased products such that the total purchase quantity obtained by summing the initial purchase quantity and the regular purchase quantity of regularly purchased products is maximized. The purchase quantity estimation device for regularly purchased products according to Claim 1.

3. The estimation unit predicts the initial purchase quantity of regularly purchased products and predicts the regular purchase quantity of regularly purchased products such that the total purchase quantity obtained by summing the initial purchase quantity and the regular purchase quantity of regularly purchased products becomes the target purchase quantity. The order quantity estimation device for regularly purchased products according to Claim 1.

4. An actual value acquisition unit that acquires the actual value of the initial purchase quantity of regularly purchased products and the actual value of the publication volume for each advertising publication medium for which advertisements have been published for regularly purchased products; A prediction model generation unit that generates a prediction model for estimating the initial purchase quantity of regularly purchased products from the publication volume for each advertising publication medium using the actual value of the initial purchase quantity and the actual value of the publication volume. The purchase quantity estimation device for regularly purchased products according to Claim 1, comprising:

5. The survival rate acquisition unit includes: An invalid month acquisition unit that acquires the number of invalid months for determining that a customer who has made an initial purchase of a regularly purchased product has stopped regular purchases; A regular purchase stop determination unit that compares the number of invalid months and the number of months elapsed from the final purchase date of the regularly purchased product for each customer who has made an initial purchase of the regularly purchased product to determine that the customer who has made an initial purchase of the regularly purchased product has stopped regular purchases; A survival period calculation unit that calculates the survival period from the initial purchase date to the final purchase date of the regularly purchased product for each customer who has made an initial purchase of the regularly purchased product; A survival rate calculation unit that calculates the survival rate at which customers continue to purchase regularly purchased products using the survival period for each customer who has made an initial purchase of the regularly purchased product. The purchase quantity estimation device for regularly purchased products according to Claim 1, including:

6. A purchase quantity estimation program for regularly purchased products that causes a computer to execute the following processes to estimate the purchase quantity of regularly purchased products: An invalid month number acquisition process for acquiring the invalid month number for determining that a customer who has made an initial purchase of a regularly purchased product has stopped the regular purchase, A regular purchase stop determination process for comparing, for each customer who has made an initial purchase of a regularly purchased product, the invalid month number with the number of months elapsed since the final purchase date of the regularly purchased product to determine that the customer who has made an initial purchase of the regularly purchased product has stopped the regular purchase, A survival period calculation process for calculating, for each customer who has made an initial purchase of a regularly purchased product, the survival period from the initial purchase date of the regularly purchased product to the final purchase date of the regularly purchased product, A survival rate calculation process for calculating the survival rate at which a customer continues to purchase a regularly purchased product using the survival period of each customer who has made an initial purchase of the regularly purchased product, An initial purchase quantity prediction value acquisition process for acquiring a predicted value of the initial purchase quantity of a regularly purchased product, An estimation process for estimating the predicted value of the regular purchase quantity of a regularly purchased product or / and the predicted value of the total purchase quantity obtained by summing the initial purchase quantity and the regular purchase quantity of the regularly purchased product using the survival rate and the predicted value of the initial purchase quantity, A purchase quantity estimation program for regularly purchased products including the above.

7. A purchase quantity estimation device for regularly purchased products that estimates the purchase quantity of regularly purchased products, An invalid month number acquisition unit for acquiring the invalid month number for determining that a customer who has made an initial purchase of a regularly purchased product has stopped the regular purchase, A regular purchase stop determination unit for comparing, for each customer who has made an initial purchase of a regularly purchased product, the invalid month number with the number of months elapsed since the final purchase date of the regularly purchased product to determine that the customer who has made an initial purchase of the regularly purchased product has stopped the regular purchase, A survival period calculation unit for calculating, for each customer who has made an initial purchase of a regularly purchased product, the survival period from the initial purchase date of the regularly purchased product to the final purchase date of the regularly purchased product, A survival rate calculation unit for calculating the survival rate at which a customer continues to purchase a regularly purchased product using the survival period of each customer who has made an initial purchase of the regularly purchased product, An initial purchase quantity prediction value acquisition unit for acquiring a predicted value of the initial purchase quantity of a regularly purchased product, An estimation unit for estimating the predicted value of the regular purchase quantity of a regularly purchased product or / and the predicted value of the total purchase quantity obtained by summing the initial purchase quantity and the regular purchase quantity of the regularly purchased product using the survival rate and the predicted value of the initial purchase quantity, A purchase quantity estimation device for regularly purchased products provided with the above.

8. The invalid month number acquisition unit, Is configured to include an invalid month number reception unit that receives the invalid month number on the display screen of the display device of the computer, The purchase quantity estimation device for regularly purchased products according to claim 7.

9. The estimation unit, Using a prediction model that estimates the first purchase quantity of a regularly purchased product from the publication volume for each advertising publication medium that has published an advertisement for regularly purchasing the product, the first purchase quantity of the regularly purchased product is predicted, and the regular purchase quantity of the regularly purchased product is predicted using the predicted first purchase quantity and the survival rate. The purchase quantity estimation device for regularly purchased products according to claim 7.

Images