Estimated value calculation method, estimated value calculation system, and client device

Abstract

To provide an estimated value calculation method, estimated value calculation system, and client device that can calculate with higher accuracy an estimated value of unrecorded data that is not included in sales-related statistical data.SOLUTION: An estimated value calculation method, which is a method for calculating an estimated value of unrecorded data of an object item in sales-related statistical data, includes: acquiring recorded data of the object item from the sales-related statistical data; acquiring first purchase data corresponding to the recorded data, which is purchase data obtained by collecting purchase records of purchasers; generating a model showing a relationship between the recorded data and the first purchase data; acquiring second purchase data corresponding to the unrecorded data, which is the purchase data; and calculating the estimated value of the unrecorded data from the second purchase data based on the model.SELECTED DRAWING: Figure 13

Description

[Technical field]

[0001] SUMMARY OF THE DISCLOSURE The present invention relates to an estimate calculation method, an estimate calculation system, and a client device. [Background technology]

[0002] The Household Income and Expenditure Survey (Household Income and Expenditure Edition) provided by the Statistics Bureau of the Ministry of Internal Affairs and Communications is a statistical survey of household income and expenditure conducted monthly on approximately 9,000 households nationwide selected based on a set statistical sampling method. Hereafter, data related to expenditures from the Household Income and Expenditure Survey (Household Income and Expenditure Edition) will be referred to as "household expenditure data." Household expenditure data is widely used by government and companies in various analyses of consumption as an indicator of personal consumption.

[0003] The household expenditure data may be data collected by representative region, data collected by month instead of by day, or data with a certain time lag that does not include the latest data. Therefore, the household expenditure data may not include data that the user desires. Therefore, a method for calculating an estimate of unrecorded data that is not included in the household expenditure data has been devised (for example, Patent Document 1).

[0004] By using past published household expenditure data, it is possible to calculate estimates of unrecorded data (such as unpublished data) that is not included in the current household expenditure data. It is desirable to efficiently learn from past published household expenditure data to calculate estimates with higher accuracy, and to calculate estimates with higher accuracy by taking into account current events occurring in the present. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] JP 2015-088183 A Summary of the Invention [Problem to be solved by the invention]

[0006] The problem to be solved by the present invention is to provide an estimated value calculation method, an estimated value calculation system, and a client device that can calculate, with higher accuracy, estimated values ​​of unrecorded data that is not included in buying and selling related statistical data, such as household expenditure data. [Means for solving the problem]

[0007] An embodiment of an estimated value calculation method is a method for calculating an estimated value of unrecorded data of a target item in sales-related statistical data, which includes obtaining recorded data for the target item from the sales-related statistical data, obtaining first purchasing data corresponding to the recorded data which is purchasing data collected from purchasers' purchasing records, generating a model showing a relationship between the recorded data and the first purchasing data, obtaining second purchasing data corresponding to the unrecorded data which is purchasing data, and calculating the estimated value of the unrecorded data from the second purchasing data based on the model. [Brief description of the drawings]

[0008] [Figure 1] FIG. 1 is a configuration diagram of an estimated value calculation system according to a first embodiment. [Diagram 2] FIG. 2 is a diagram showing a specific example of the estimated value calculation system. [Diagram 3] FIG. 2 is a functional block diagram of the estimated value calculation system. [Figure 4] FIG. 2 is a functional block diagram of a user interface unit and a recording unit of the estimated value calculation system. [Diagram 5] FIG. 11 is a diagram showing an example of common classification data. [Figure 6] FIG. 4 is a diagram showing an example of collected purchase data. [Figure 7] FIG. 13 is a diagram showing an example of processed, collected purchasing data. [Figure 8] FIG. 4 is a diagram showing an example of household expenditure data. [Figure 9] FIG. 13 is a diagram showing an example of the processed household expenditure data by month. [Figure 10]4 is a flowchart of the learning phase of the estimation value calculation system. [Figure 11] FIG. 2 is a diagram showing an example of daily household expenditure data. [Figure 12] FIG. 13 is a diagram showing an example of the processed household expenditure data by day. [Figure 13] FIG. 11 is a diagram showing an example of recorded data. [Figure 14] FIG. 13 is a diagram showing the same processed and collected purchasing data of a corresponding product, to which an amount equivalent to the standard price has been added. [Figure 15] FIG. 13 is a diagram showing daily processed and collected purchasing data of corresponding products, in which the price level is added as first index data. [Figure 16] FIG. 11 is a diagram showing the operation of the calculation unit in step S160. [Figure 17] 4 is a flowchart of the estimation phase of the estimation value calculation system. [Figure 18] FIG. 13 is a diagram showing the operation of the calculation unit in step S230. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] Hereinafter, an estimated value calculation system, an estimated value calculation method, and a client device according to embodiments will be described with reference to the drawings.

[0010] (First embodiment) [Estimated value calculation system 100] FIG. 1 is a configuration diagram of an estimated value calculation system 100 according to the first embodiment. The estimated value calculation system 100 is a system that calculates an estimated value of data that is not included in data on expenditures (hereinafter also referred to as "household expenditure data") among data from the Household Income and Expenditure Survey (Household Income and Expenditure Edition) provided by the Statistics Bureau of the Ministry of Internal Affairs and Communications.

[0011] The estimated value calculation system 100 is composed of one or more servers. The basic infrastructure of the estimated value calculation system 100 is constructed using a known data distributed processing system. The functions of the estimated value calculation system 100 are realized by a program executed in the processor of the server.

[0012] FIG. 2 is a diagram showing a specific example of the estimated value calculation system 100. As shown in FIG. 2 includes servers SV1, SV2, SV3, and SV4. The servers SV1, SV2, SV3, and SV4 include hardware such as a processor such as a CPU, a memory, a recording medium such as a hard disk or a flash memory, and a communication device.

[0013] The servers SV1, SV2, SV3, and SV4 may be servers with different types of OS, software, etc. The servers SV1, SV2, SV3, and SV4 may be operated by different operating entities. The servers SV1, SV2, SV3, and SV4 may be installed in different countries or different regions.

[0014] All of the hardware and software resources of the servers SV1 and SV2 are occupied by the estimate calculation system 100. On the other hand, only a portion of the hardware and software resources of the servers SV3 and SV4 are provided to the estimate calculation system 100. In other words, the estimate calculation system 100 may be a system that combines on-site servers such as the servers SV1 and SV2 with cloud servers such as the servers SV3 and SV4 that provide IaaS (Infrastructure as a Service), PaaS (Platform as a Service), or SaaS (Software as a Service). In the following description, "server" includes any of the above servers.

[0015] The functional blocks of the estimated value calculation system 100 described below (such as a data acquisition unit 1, a user interface unit 2, a recording unit 3, a calculation unit 4, and a data output unit 5) are realized by the operation of servers (for example, servers SV1, SV2, SV3, and SV4 illustrated in FIG. 2) that are distributed and arranged via a network NW. Note that the functional blocks of the estimated value calculation system 100 may be realized by the operation of a single server.

[0016] 1, the estimated value calculation system 100 is connected to a data device 200 managed by a data provider (DP) of the Statistics Bureau of the Ministry of Internal Affairs and Communications, a data device 250 managed by the Statistics Bureau of the Ministry of Internal Affairs and Communications, a terminal 300 (hereinafter referred to as a personal terminal 300) such as a smartphone, a tablet terminal, or a personal computer owned by an individual user U, a management device 400 owned by an administrator A of the estimated value calculation system 100, and a service providing device (client device) 500 owned by a service provider (SP) via a network NW. The data device 200, the data device 250, the management device 400, and the service providing device 500 may be a system composed of one or more servers.

[0017] The network NW may be a wide area network (WAN) such as the Internet, a private network (LAN) in a building where the estimation value calculation system 100 is installed, or a combination of these. The network NW may be either wired or wireless. Other devices such as an access point may be involved in the communication between the devices.

[0018] FIG. 3 is a functional block diagram of the estimated value calculation system 100. The estimated value calculation system 100 includes a data acquisition unit 1, a user interface unit 2, a recording unit 3, a calculation unit 4, and a data output unit 5.

[0019] [Data Acquisition Section 1] The data acquisition unit 1 acquires collected purchase data 33 including purchase records of individual user U from the data device 200 of the data provider DP and records it in the recording unit 3. The data acquisition unit 1 can acquire collected purchase data 33 including purchase records of individual user U from a first data provider DB1 and a second data provider DB2, which are different data providers DP.

[0020] Specifically, the data acquisition unit 1 acquires collected purchase data 33 including purchase records of individual users U from the data device 200 of the data provider DP based on a contract with the data provider DP, and records the data in the recording unit 3. The data acquisition unit 1 may, for example, periodically acquire data in bulk from the data device 200 of the data provider DP (bulk type). The data acquisition unit 1 may also, for example, acquire data in real time from a device connected via an API (Application Programming Interface) (streaming type).

[0021] The data device 200 of the data provider DP acquires purchase records such as electronic receipts and household account books of the individual user U from the personal terminal 300 based on a contract with the individual user U or the like.

[0022] The data acquisition unit 1 acquires household expenditure data 35 from a data device 250 managed by the Statistics Bureau of the Ministry of Internal Affairs and Communications, and records the data in the recording unit 3. The data acquisition unit 1 periodically acquires household expenditure data 35 that is provided periodically.

[0023] [User Interface Part 2] FIG. 4 is a functional block diagram of the user interface unit 2 and the recording unit 3. The user interface unit 2 is an interface connected to a management device 400 owned by an administrator A of the estimated value calculation system 100. Specifically, the user interface unit 2 includes an API that connects the management device 400 and the recording unit 3. The user interface unit 2 has an interface 21 for administrator.

[0024] The administrator interface 21 includes an API that connects the management device 400 owned by the administrator A and the recording unit 3. The management device 400 can be connected to the system setting information recording unit 31 of the recording unit 3 via the administrator interface 21.

[0025] [Record 3] The recording unit 3 records purchase information such as purchase data, system setting information, etc. The recording unit 3 may be distributed to the servers SV1, SV2, SV3, and SV4. The recording unit 3 has a system setting information recording unit 31 and a purchase information recording unit 32.

[0026] The system setting information recording unit 31 records system setting information 311 which is the system setting of the estimated value calculation system 100. The system setting information 311 is updated by the management device 400 connected via the manager interface 21.

[0027] The purchasing information recording unit 32 records collected purchasing data 33, processed collected purchasing data 34, household expenditure data 35, processed household expenditure data 36, ​​common classification data 37, causal data 38, and trained model 39.

[0028] FIG. 5 is a diagram showing an example of the common classification data 37. As shown in FIG. The common classification data 37 is an item classification commonly used in the estimated value calculation system 100. The common classification data 37 is information that associates the classification of the household expenditure data (classification information 354 described later) and the classification of the collected purchase data (product classification information 335 described later) with respect to the common item classification information 371.

[0029] The common item classification information 371 is information that identifies an item, such as a common item classification.

[0030] The classification of the collected purchase data in the common classification data 37 (product classification information 335 described later) may include a "classification based on product name" that is an appropriate product classification determined from the product name. The classification of the collected purchase data (product classification information 335 described later) can be uniquely associated with the common item classification information 371.

[0031] FIG. 6 is a diagram showing an example of the collected purchase data 33. As shown in FIG. The collected purchase data 33 is information in which date and time information 331, user information 332, store information 333, product information 334, product classification information 335, and purchase information 336 are associated with each purchased product.

[0032] The date and time information 331 is information for identifying the date and time when the product was purchased, such as a date.

[0033] The user information 332 is information for identifying an individual user U who has purchased a product, such as a user ID.

[0034] The store information 333 is information for identifying the store where the product was purchased, such as a store ID.

[0035] Product information 334 is information that identifies the purchased product, such as a product ID and a product name. Product information 334 for a product that does not have a product ID assigned, such as fresh food such as Chinese cabbage, does not include a product ID and includes only the product name.

[0036] Product classification information 335 indicates the classification of purchased products. Product classification in product classification information 335 is hierarchical, and in the example shown in FIG. 6, it is hierarchical into three levels (major product classification, medium product classification, minor product classification). The major product classification is the highest classification with the greatest granularity among the three levels. The minor product classification is the lowest classification with the smallest granularity among the three levels. Product classification information 335 for products whose product classification is not common between stores (non-common products), such as the product with the product name "green onion" in FIG. 6, does not include the product classification.

[0037] The purchase information 336 is information related to the purchase of goods, such as the number of goods purchased and the unit price.

[0038] FIG. 7 is a diagram showing an example of the processing, collection, and purchase data 34. As shown in FIG. The processed collected purchase data 34 is data obtained by processing the product classification information 335 of the collected purchase data 33 based on the common classification data 37. For example, the processed collected purchase data 34 has the common item classification information 371 corresponding to the product classification information 335 added thereto (the product classification information 335 is omitted in FIG. 7). For a product with the product name "Chinese cabbage 1 / 2", it is not possible to determine whether the corresponding common item classification is "Chinese cabbage" or "Green onion" based on the major product classification, medium product classification, and minor product classification. In this case, the corresponding common item classification of the product with the product name "Chinese cabbage 1 / 2" is determined to be "Chinese cabbage" using the "Classification from product name" included in the common classification data 37. Also, for a product without product classification information 335, such as the product with the product name "Green onion" in FIG. 6, the corresponding common item classification is determined to be "Green onion" using the "Classification from product name" included in the common classification data 37.

[0039] The standardization of item classification for the product classification information 335 based on the common classification data 37 may be performed based on a rule base, or may be performed using a trained model generated by machine learning.

[0040] FIG. 8 is a diagram showing an example of the household expenditure data 35. As shown in FIG. The household expenditure data 35 is data on expenditures from the household survey (household income and expenditure edition) provided by the Statistics Bureau of the Ministry of Internal Affairs and Communications. For example, the survey results of households with two or more people in the household expenditure data are mainly data summarized in monthly income and expenditure amounts (including purchase quantities by item) per household for each region, household, and income category. The household expenditure data 35 may be provided by other survey organizations as long as it contains equivalent data. The household expenditure data 35 is information in which date and time information 351, region information 352, classification information 354, household information 355, and expenditure information 356 are associated with each item. The household expenditure data 35 is an example of sales-related statistical data. Here, "sales-related statistical data" includes various expenditure-related statistical data of consumers or businesses (including retailers, wholesalers, and other service industry businesses), including household expenditure data from the household expenditure survey, and sales-related statistical data of businesses, such as the "Service Industry Trend Survey" from the Statistics Bureau of the Ministry of Internal Affairs and Communications and the "Commercial Activity Statistics Survey" from the Minister's Secretariat of the Ministry of Economy, Trade and Industry.

[0041] The date and time information 351 is information that specifies the date and time when the item was purchased, such as the year, month, and date.

[0042] The area information 352 is information for identifying the area where the item was purchased, and is, for example, a name indicating the area, such as "nationwide," "Kanto," or "Kansai."

[0043] The classification information 354 indicates the classification of the purchased item. The classification information 354 is based on the product classification defined in the Annual Household Survey provided by the Statistics Bureau of the Ministry of Internal Affairs and Communications. The product classification in the classification information 354 is hierarchically organized into four levels (major classification, medium classification, minor classification, item classification) as shown in FIG. 8. The product classification in the classification information 354 of the household expenditure data 35 differs from the product classification in the product classification information 335 of the collected purchase data 33. The major classification is the upper classification with the greatest granularity among the four levels. The item classification is the lower classification with the smallest granularity among the four levels.

[0044] Household information 355 is information about the number of people in the household that made the purchase, for example, "single" or "two or more people."

[0045] Expenditure information 356 is expenditure corresponding to classification information 354 per household, and includes item expenditure (yen / household) and purpose expenditure (yen / household). Item expenditure (yen / household) and purpose expenditure (yen / household) are set out in the Annual Household Income and Expenditure Survey provided by the Statistics Bureau of the Ministry of Internal Affairs and Communications. Specifically, item expenditure (yen / household) is the total expenditure corresponding to the four-level classification described above that is not dependent on the purpose (purpose) of expenditure, and includes food expenses within one's own household as well as gifts to other households. Purpose expenditure (yen / household) is the total expenditure corresponding to the four-level classification described above that is limited to consumption within one's own household.

[0046] FIG. 9 is a diagram showing an example of the monthly processed household expenditure data 36. The processed household expenditure data 36 is data obtained by processing the classification information 354 of the household expenditure data 35 based on the common classification data 37. For example, common item classification information 371 corresponding to the classification information 354 is added to the processed household expenditure data 36. Also, the item expenditures of the expenditure information 356 in the processed household expenditure data 36 are supplemented with estimated values. Also, national item expenditures, which are the expenditures of all households nationwide, are added. The method of calculating the item expenditures and the estimated values ​​of the national item expenditures will be described later.

[0047] The standardization of item classification for the classification information 354 based on the common classification data 37 may be performed based on a rule base, or may be performed using a trained model generated by machine learning.

[0048] The causal data 38 is data on external factors that affect sales and the number of customers visiting the store. The causal data 38 includes, for example, calendar information, weather information, and local event information.

[0049] The trained model 39 is a model that has learned the relationship between the processed collected purchasing data 34 (or the collected purchasing data 33), the processed household expenditure data 36 (household expenditure data 35), and the causal data 38. As described below, the trained model 39 is a model generated by machine learning. Note that the trained model 39 may be a model generated by a simple regression method such as regression analysis or simple averaging.

[0050] The calculation unit 4 calculates an estimate of unrecorded data S2 not included in the household expenditure data 35 based on the processed household expenditure data 36 and the causal data 38, for example, to generate the processed household expenditure data 36.

[0051] The data output unit 5 outputs the estimated values ​​of the unrecorded data S2 calculated by the calculation unit 4 in response to a data access request from the service providing device (client device) 500 of the service provider SP.

[0052] [Operation of the estimated value calculation system 100] Next, a description will be given of the operation of the estimated value calculation system 100. Fig. 10 is a flowchart of the learning phase of the estimated value calculation system 100. Note that the order of execution of multiple steps in the flowchart that have no dependency relationship may be changed.

[0053] <Step S100> In step S100, the calculation unit 4 generates processed household expenditure data 36 from the household expenditure data 35 recorded in the recording unit 3. As shown in FIG. 9, the processed household expenditure data 36 is supplemented with common item classification information 371 and with the item expenditure and nationwide item expenditure of the expenditure information 356. The calculation unit 4 records the generated processed household expenditure data 36 in the recording unit 3.

[0054] The estimated value EV1 shown in Figure 9 is the expenditure (yen / household) for all items in the "single-person" household type and the "fresh vegetables" medium category. The estimated value EV1 is calculated using Equation 1.

[0055] Values ​​V1, V2, and V3 in formula 1 are data recorded in household expenditure data 35 as shown in Figure 8. Value V1 is the expenditure (yen / household) for all items in the "single person" household type and the "fresh vegetables" medium category. Value V2 is the expenditure (yen / household) for all items in the "two or more people" household type and the "fresh vegetables" medium category. Value V3 is the expenditure (yen / household) for all items in the "two or more people" household type and the "fresh vegetables" medium category.

[0056]

number

[0057] The estimated value EV2 shown in Figure 9 is the item expenditure (yen / household) for the item whose household type is "single person", whose subcategory is "Chinese cabbage", and whose item classification is "Chinese cabbage". The estimated value EV2 is calculated using Equation 2.

[0058] The value V4 in the formula 2 is data recorded in the household expenditure data 35 as shown in Figure 8. The value V4 is the item expenditure (yen / household) for the item whose household type is "two or more people", whose subcategory is "Chinese cabbage", and whose item classification is "Chinese cabbage".

[0059]

number

[0060] The estimated value EV3 shown in Figure 9 is the national item expenditure (yen) for the item whose household type is "all households" and whose common item classification is "Chinese cabbage." The estimated value EV3 is calculated using Equation 3.

[0061] The number of single-person households and the number of two-or-more-person households in Equation 3 are data provided by national survey results and municipalities, etc. Here, the number of single-person households is set to 21,151,042 people, and the number of two-or-more-person households is set to 34,553,907 people.

[0062]

number

[0063] FIG. 11 is a diagram showing an example of the daily household expenditure data 35. As shown in FIG. Here, the household expenditure data 35 includes daily item expenditures (VDi) for household types of "two or more people." i is the date. It is assumed that the household expenditure data 35 does not include daily item expenditures for household types of "single-person" and "all households." However, the household expenditure data 35 may include daily item expenditures for household types of "single-person" and "all households."

[0064] FIG. 12 is a diagram showing an example of the processed household expenditure data 36 by day. The estimated daily item expenditure, EV4, when the household type is “all households” is calculated using Equation 4. ΣVDi is the monthly total of VDi.

[0065]

number

[0066] <Step S110> In step S110, the calculation unit 4 acquires the recorded data S1 of the target item from the processed household expenditure data 36 recorded in the recording unit 3. In this embodiment, the recorded data S1 is daily data, but it may be weekly or monthly data. The recorded data S1 may be limited by region, period, purchaser attributes, etc. The "target item" is an item included in the household expenditure data 35, and is an item for which an estimated value of the unrecorded data S2 described below is calculated. The calculation unit 4 may acquire the recorded data S1 of the target item from the household expenditure data 35 recorded in the recording unit 3.

[0067] FIG. 13 is a diagram showing an example of the recorded data S1 etc. The recorded data S1 illustrated in Fig. 13 includes data from a specific time period to the latest recorded data. The unrecorded data S2 estimated hereinafter is past data that has not yet been included in the processed household expenditure data 36 (household expenditure data 35). The recorded data S1 is not limited to the example illustrated in Fig. 13.

[0068] <Step S120> In step S120, the calculation unit 4 generates the processed, collected, purchased data 34 from the collected purchased data 33 recorded in the recording unit 3. The processed, collected, purchased data 34 has common item classification information 371 added thereto, as shown in FIG. 7. The calculation unit 4 records the generated processed, collected, purchased data 34 in the recording unit 3.

[0069] <Step S130> In step S130, the calculation unit 4 acquires the first purchase data P1 from the processed collected purchase data 34 recorded in the recording unit 3. The first purchase data P1 is data corresponding to the recorded data S1 of the target item. Specifically, the first purchase data P1 is data of a product (hereinafter also referred to as a "corresponding product") that has the same or similar common item classification as the target item, and is data equivalent to the recorded data S1 in terms of region, period, purchaser attributes, etc. In this embodiment, the first purchase data P1 is daily data corresponding to the recorded data S1. If the recorded data S1 is weekly data, the first purchase data P1 is also weekly data. The calculation unit 4 may acquire the first purchase data P1 from the collected purchase data 33 recorded in the recording unit 3.

[0070] <Step S140> In step S140, the calculation unit 4 generates first index data, which is index data, from the first purchasing data P1. The index data is data obtained by standardizing the price or purchase amount of the product in the purchasing data based on a predetermined rule. The index data includes at least one of a price level and an opportunity-considered purchasing amount, which will be described later. Note that the generation of the first index data is not essential.

[0071] FIG. 14 is a diagram showing the processed, collected, and purchased data 34 of the corresponding product to which the standard price equivalent has been added. To explain the price levels, the standard price equivalent will first be explained. The standard price equivalent is the standard price of the corresponding product that has not been discounted. If the standard price equivalent is unknown, the maximum purchase amount (unit price) of the corresponding product in the first purchasing data P1 is taken as the standard price equivalent that has not been discounted. In the example shown in FIG. 14, the standard price equivalent for a product with the product name "Chinese Cabbage 1 / 4" during a specified period is 120 yen. The standard price equivalent for a product with the product name "Chinese Cabbage 1 / 2" during a specified period is 250 yen.

[0072] FIG. 15 is a diagram showing the daily processed and collected purchase data 34 of the corresponding product, in which the price level is added as the first index data. The price level is the ratio of the accumulated value of the purchase amount (unit price) to the accumulated value of the standard price equivalent amount of the corresponding product (product belonging to a common item classification) in the first purchase data P1. In the example shown in FIG. 15, the total purchase amount of the target product (product whose common item classification is "Chinese cabbage") on the date "2021 / 2 / 1" is 985 yen (110 yen + 250 yen + 110 yen + 105 yen + 100 yen + 70 yen + 240 yen). On the other hand, the total standard price equivalent amount is 1100 yen (120 yen + 250 yen + 120 yen + 120 yen + 120 yen + 120 yen + 250 yen). Therefore, the price level of the target product (the product whose common item classification is "Chinese cabbage") on the date "2021 / 2 / 1" is "0.895".

[0073] Opportunity-considered purchase is the purchase amount per purchase opportunity of the corresponding product in the first purchase data P1, specifically, the purchase amount per person who had the opportunity to purchase the in-store stock. If the number of people who had the opportunity to purchase the in-store stock (in-store stock contacts) is unknown, the number of people who purchase the products included in the classification (major product classification, medium product classification, or minor product classification) to which the corresponding product belongs may be estimated as the in-store stock contacts to calculate opportunity-considered purchase.

[0074] <Step S150> In step S150, the calculation unit 4 extracts the first causal data C1 corresponding to the first purchase data P1 from the causal data 38 recorded in the recording unit 3. Specifically, the first causal data C1 is causal data having the same region, period, etc. as the first purchase data P1. Note that the extraction of the first causal data C1 is not essential.

[0075] <Step S160> In step S160, the calculation unit 4 generates a trained model 39 indicating the relationship between the collected data S1, the first purchase data P1, and the first causal data C1. The calculation unit 4 records the generated trained model 39 in the recording unit 3.

[0076] FIG. 16 is a diagram showing the operation of the calculation unit 4 in step S160. The trained model 39 is trained so that it can output a value that is closer to the recorded data S1 than the first purchasing data P1 and the first causal data C1 that are input.

[0077] The first index data is treated as additional data to the first purchase data P1. By using the first index data, the calculation unit 4 can improve the accuracy of inference by machine learning using the trained model 39. Note that the first purchase data P1 input to the trained model 39 may be only the first index data.

[0078] When the first causal data C1 is not present, the calculation unit 4 generates a trained model 39 indicating the relationship between the recorded data S1 and the first purchase data P1.

[0079] 17 is a flowchart of the estimation phase of the estimated value calculation system 100. Note that the order of execution of multiple steps in the flowchart that have no dependency on each other may be changed.

[0080] <Step S200> In step S200, the calculation unit 4 acquires second purchasing data P2 from the processed collected purchasing data 34 recorded in the recording unit 3. The second purchasing data P2 is data corresponding to the unrecorded data S2 of the target item. Specifically, the second purchasing data P2 is data of a product that is the same as or similar to the target item, and is data equivalent to the unrecorded data S2 in terms of region, period, purchaser attributes, etc. The calculation unit 4 may acquire the second purchasing data P2 from the collected purchasing data 33 recorded in the recording unit 3.

[0081] <Step S210> In step S210, the calculation unit 4 generates second index data, which is index data, from the second purchase data P2. The index data includes at least one of a price level and an opportunity-considered purchase amount. Note that the generation of the second index data is not essential.

[0082] <Step S220> In step S220, the calculation unit 4 extracts second causal data C2 corresponding to the second purchasing data P2 from the causal data 38 recorded in the recording unit 3. Specifically, the second causal data C2 is causal data equivalent to the second purchasing data P2 in terms of region, period, etc. Note that if the first causal data C1 is not extracted, the second causal data C2 is not essential.

[0083] <Step S230> In step S230, the calculation unit 4 calculates an estimate of the unrecorded data S2 from the second purchase data P2 based on the trained model 39.

[0084] FIG. 18 is a diagram showing the operation of the calculation unit 4 in step S230. The trained model 39 outputs an estimated value of the unrecorded data S2 for the input second purchasing data P2 and second causal data C2.

[0085] <Step S240> In step S240, the calculation unit 4 uses a known BI (Business Intelligence) tool to visualize the processed household expenditure data 36 (household expenditure data 35) to which the unrecorded data S2 has been added.

[0086] According to the estimated value calculation system 100, the estimated value of the unrecorded data S2 not included in the household expenditure data 35 can be calculated with high accuracy by using the collected purchase data 33, which is another data. In addition, the estimated value of the unrecorded data S2 can be calculated with high accuracy by standardizing the item classification of the collected purchase data 33 and the household expenditure data 35 using the common classification data 37. In addition, the accuracy of inference based on a model showing the relationship between the collected purchase data 33 and the household expenditure data 35 can be improved by using index data (price level, opportunity-considered purchase amount) calculated from the collected purchase data 33. In addition, the estimated value of the unrecorded data S2 can be calculated with high accuracy by taking into account current events occurring at the moment by using the causal data 38. In addition, although the present embodiment has described the calculation of estimated values ​​of item expenditures and national item expenditures, the expenditure on purpose (yen / household) for an item classification (e.g., "Chinese cabbage") may be calculated and supplemented by using the ratio of item expenditures to expenditure on purpose for a classification that includes the item classification (e.g., "Chinese cabbage") (e.g., "fresh vegetables," a subclassification that includes "Chinese cabbage"). In addition, by utilizing the collected purchasing data 33, standardizing item classification, utilizing index data calculated from the collected purchasing data 33, and utilizing causal data as shown in this embodiment, it is possible to accurately calculate estimates of unrecorded data such as "sales-related statistical data" other than the household expenditure data of the household budget survey.

[0087] Although some embodiments of the present invention have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their modifications are included in the scope of the invention and its equivalents described in the claims, as well as in the scope and spirit of the invention. [Explanation of symbols]

[0088] 100: Estimated value calculation system, 200: Data device, 300: Personal terminal, 1: Data acquisition section, 2: User interface section, 3: Recording section, 4: Calculation section, 5: Data output section

Claims

1. A method for calculating an estimate of unrecorded data for a target item in trade-related statistical data, comprising: Acquire recorded data of the target item from the trading-related statistical data; Acquire first purchase data corresponding to the collected purchase records of purchasers, generating a model indicating a relationship between the recorded data and the first purchase data; Acquire second purchasing data corresponding to the unrecorded data, calculating the estimated value of the unrecorded data from the second purchase data based on the model; Method of calculating estimates.

2. generating the model using first index data, which is index data calculated from the first purchase data; calculating the estimated value using second index data, which is the index data calculated from the second purchase data; The method of claim 1 .

3. The index data is a ratio of the purchase amount of the target item to the standard price equivalent of the target item. The method for calculating an estimated value according to claim 2.

4. The standard price equivalent is the maximum purchase amount of the target item during a specified period. The method for calculating an estimated value according to claim 3.

5. The index data is a purchase amount of the target item per contact person who had an opportunity to purchase the target item in store inventory. The method for calculating an estimated value according to claim 2.

6. generating the model using causal data corresponding to the first purchasing data; calculating the estimated value using the causal data corresponding to the second purchase data; The method of claim 1 .

7. The model is a model that learns the relationship between the recorded data and the first purchase data by machine learning. The method of claim 1 .

8. The model is a model calculated by performing regression analysis on the relationship between the recorded data and the first purchase data. The method of claim 1 .

9. Creating processing trade-related statistical data by changing the classification of the trade-related statistical data based on the common classification; The recorded data is obtained from the processing trade related statistical data, Creating processed and collected purchasing data by changing the classification of the purchasing data based on the common classification; The first purchasing data and the second purchasing data are obtained from the processed and collected purchasing data. The method of claim 1 .

10. A recorder for recording buying and selling related statistical data and purchasing data obtained by collecting purchasing records of purchasers; a calculation unit that calculates an estimated value of unrecorded data of a target item in the trading-related statistical data based on the trading-related statistical data and the purchase data; Equipped with The calculation unit is Acquire recorded data of the target item from the trading-related statistical data; Acquire first purchasing data corresponding to the recorded data, generating a model indicating a relationship between the recorded data and the first purchase data; Acquire second purchasing data corresponding to the unrecorded data, calculating the estimated value of the unrecorded data from the second purchase data based on the model; Estimate calculation system.

11. The calculation unit is generating the model using first index data, which is index data calculated from the first purchase data; calculating the estimated value using second index data, which is the index data calculated from the second purchase data; The estimation system according to claim 10.

12. The index data is a ratio of the purchase amount of the target item to the standard price equivalent of the target item. The estimate calculation system according to claim 11.

13. A method for obtaining an estimate of unrecorded data for a target item in trade-related statistical data, comprising: The unrecorded data is Acquire recorded data of the target item from the trading-related statistical data; Acquire first purchase data corresponding to the collected purchase records of purchasers, generating a model indicating a relationship between the recorded data and the first purchase data; Acquire second purchasing data corresponding to the unrecorded data, the estimated value of the unrecorded data is calculated from the second purchase data based on the model; How to obtain estimates.

14. A client device that acquires an estimated value of unrecorded data of a target item in trading-related statistical data from a server device, The server device includes: A recorder for recording buying and selling related statistical data and purchasing data obtained by collecting purchasing records of purchasers; a calculation unit that calculates an estimated value of unrecorded data of a target item in the trading-related statistical data based on the trading-related statistical data and the purchase data; Equipped with The calculation unit is Acquire recorded data of the target item from the trading-related statistical data; Acquire first purchasing data corresponding to the recorded data, generating a model indicating a relationship between the recorded data and the first purchase data; Acquire second purchasing data corresponding to the unrecorded data, calculating the estimated value of the unrecorded data from the second purchase data based on the model; Client device.

Images