JP6748769B1 - Customer estimation device and customer estimation method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a customer estimation device and a customer estimation method capable of accurately estimating potential customers of items to be indexed. SOLUTION: A customer estimation device 1 has an indexing item acquisition unit 2 for acquiring indexed items, a customer-owned item acquisition unit 3 for acquiring customer-owned items, and an item for associating a customer-owned item with a source data item. Based on a predetermined rule, the corresponding unit 4, the classification item creation unit 5 that classifies the customer-owned items according to the contents of the indexed items and creates the classification items, and the classification items classified by the classification item creation unit 5 A score giving unit 6 for giving a score is provided. [Selection diagram] Fig. 1

Description

The present invention relates to a customer estimation device and a customer estimation method capable of accurately estimating a potential customer of an indexing item to be indexed.

2. Description of the Related Art Conventionally, there is known a system for delivering or proposing an advertisement of a product or service to a new customer by using an access log on the web.

For example, in the targeting advertisement providing system for members of Patent Document 1 below, members of the member-only site A operated by the member-only site operating company are members-only via the Internet from user terminals such as computers and smartphones. Access the site main server. The member can log in to the member-only site A by entering the member ID and password from the user terminal and browse the information providing page for the member.

Then, the user terminal passes the cookie ID of the server outside the member-only site separated from the member-only site A to the advertisement distribution server of the advertisement distribution company, so that the history in the member-only site A is leaked to the outside. Providing targeted advertising that matches the attributes of the members while preventing this.

Also, when the member accesses the website B operated by the website operating company from the user terminal via the Internet, the advertisement is delivered to the advertisement space of the website B entrusted with the advertisement delivery. (Paragraphs 0017 to 0027, FIG. 1).

Patent No. 5843983

However, in the system of Patent Document 1, since the advertisement is distributed based on the information acquired from the cookie ID, it is only possible to advertise a highly relevant product or service such as a product related to the product purchased by the user. was there. That is, it has not been possible to make a proposal for solving a matter that the user has recently become interested in or a potential problem. In order to make such a proposal, it has been premised that a concatenated identifier for identifying an individual is acquired.

The present invention has been made in view of the above problems, and it is possible to accurately estimate a potential customer of an item to be indexed from a small number of items held by the customer without acquiring a concatenated identifier for identifying an individual. An object is to provide a device and a customer estimation method.

The customer estimation device of the present invention is obtained by the indexed item acquisition means for acquiring the indexed item to be indexed, the customer owned item acquisition means for acquiring the customer owned item owned by the customer, and the customer owned item acquisition means. Acquired from the customer possession item and a survey conducted on a specific or unspecified person or the history information of the person, and at least one of the attribute, value, preference, consumption behavior, and behavior pattern of the person is reflected. an item corresponding means for corresponding the source data items in the source data, and machine learning model obtained by the machine learning the source data items corresponding to the customer ownership items as input data, depending on the contents of the indexed item , A classification rule creating unit that classifies the customer-owned items by using the machine learning model and creates classification items, and a predetermined rule using the machine learning model for the classification items created by the classification item creating unit. Score assigning means for assigning a score based on the above.

In the customer estimation device of the present invention, the indexed item acquisition means acquires, for example, the indexed items to be indexed in relation to the product to be sold or the service to be provided, and the customer possessed item acquisition means determines the gender, age from the customer. Get customer-owned items such as. Then, the item correspondence means associates the customer-owned item and the source data item with each other to link the two items, and the classification item creation means classifies the customer-owned item according to the indexed item, and the classification item. To create.

Furthermore, since the score assigning means assigns a score to each classification item based on a predetermined rule, a classification item having a high indexing item score and a classification item having a low indexing item score are selected. Can be distinguished. Therefore, it is possible to accurately estimate a potential customer who can be a target of the product or the service from these classification items.

Specifically, in the customer estimation device, the source data item corresponding to the customer-owned item ( source data in which at least one of attributes, values, tastes, consumption behaviors, and behavior patterns of a specific or unspecified person is reflected Machine learning is performed by inputting ( from the above) and a rule (machine learning model) for creating a classification item is created. Thus, when a new indexing item and customer-owned item are input, the machine learning model can be used to create classification items and assign scores based on the rules.

Further, in the customer estimation device of the present invention, it is preferable that the item handling unit adjusts to align with items of a superordinate concept when the contents of the customer-owned item and the source data item conceptually overlap.

According to this configuration, the item correspondence means compares the customer-owned item and the source data item and adjusts the item if necessary, so that the customer-owned item and the source data item are linked. When both items do not match but conceptually overlap, the item corresponding means adjusts to the item of the superordinate concept (for example, the one with coarser granularity), so it is classified from the relationship between the adjusted item and the indexed item. Can be created.

Further, in the customer estimation device of the present invention, it is preferable that the score giving means calculates a reference value for all the customers and calculates the score from a numerical value of the classification item with respect to the reference value.

According to this configuration, the score assigning unit calculates the score from the viewpoint of how large or small the numerical value of each classification item is with respect to the reference value for the entire customer. Therefore, it is possible to give a relative score to each classification item.

In addition, it is preferable that the customer estimation device of the present invention further includes a customer determination unit that determines a potential customer when the score assigned to each of the classification items exceeds a predetermined threshold value.

According to this configuration, the customer determination unit determines that the customer is a potential customer (target) regarding the indexed item when the score of each classification item exceeds the threshold value. As a result, since the target can be narrowed down as the target, an efficient approach is possible.

A customer estimation method of the present invention is a customer estimation method for estimating a target customer who is interested in a product or a service by using a computer equipped with a machine learning model, wherein the data acquisition unit of the computer is used as an index. An indexed item acquisition step of acquiring an item, the data acquisition unit acquires a customer owned item acquisition step of acquiring a customer owned item owned by the customer, and a data processing unit of the computer acquires the customer owned item acquisition step. It is acquired from the customer possessed item and a survey conducted on a specific or unspecified person or the history information of the person, and at least one of the attribute, the sense of value, the preference, the consumption behavior, and the behavior pattern of the person is An item corresponding step of making the source data item of the reflected source data correspond, a machine learning model creating step of making a machine learning model by machine learning the source data item corresponding to the customer-owned item as input data, wherein the machine learning model, depending on the content of the indexed item, and classifying the customer ownership items, a classification item creation step of creating a classification item, the machine learning model, created by the classification item creation steps And a score assigning step of assigning a score to the classification item based on a predetermined rule.

In the customer estimation method of the present invention, in the indexing item acquisition step, indexing items to be indexed in relation to the product to be sold or the service to be provided are acquired, and in the customer possessing item acquisition step, the gender and age from the customer are acquired. Get customer-owned items such as. Then, items in the corresponding step, customer ownership items and the source data item by causing the corresponding Ru done linking of both items. Further, in the machine learning model building step, after the machine learning model was created at the classification item creation steps, by classifying customer holdings item the machine learning model in accordance with the indexing item, creating a category ..

Furthermore, in the score assigning step, the machine learning model assigns a score to each classification item based on a predetermined rule, so that the score of the indexing item and the score of the indexing item are considered to be high. It is possible to distinguish the classification items that are considered to be low. Through the above steps, it is possible to accurately estimate a potential customer who can be a target of the product or service from these classification items.

The figure explaining the outline|summary of the customer estimation apparatus which concerns on 1st Embodiment of this invention. The figure explaining the example of the classification item which the customer estimation apparatus of 1st Embodiment created. Example of customer list with new items added. The figure explaining the verification result with respect to the potential customer estimated by the customer estimation device. The flowchart which estimates a potential customer using a customer estimation apparatus. The figure explaining the outline of the customer estimation apparatus which concerns on 2nd Embodiment of this invention. The figure explaining the example of the classification item which the customer estimation apparatus of 2nd Embodiment created. An example of a member list including the result of trend freak judgment. The figure explaining extraction of a trend freak.

Hereinafter, the customer estimation device of the present invention will be described in detail with reference to the drawings.
[First Embodiment]

First, with reference to FIG. 1, an outline of the customer estimation device 1 according to the first embodiment of the present invention will be described. In the first embodiment, the customer estimation device 1 is a device terminal installed in a desktop PC, a notebook PC, a tablet terminal or the like and owned by a service providing company X (hereinafter referred to as company X).

In the following, a client company of the company X, which is a supply destination Y of the customer estimation device 1 (hereinafter referred to as company Y), starts DM (direct mail) distribution and online advertisement of a newly developed drug. Therefore, a case where there is a demand for identifying a person suffering from a feeling of residual urine will be described as an example.

The company Y has a customer DB (database) 20 and holds “sex”, “age”, and “unmarried (unmarried or married)” as customer information in the customer list L. However, we do not have the direct information such as "whether there is a feeling of residual urine" which is essential.

In such a case, the company Y has an indexing item to be indexed as "whether or not to suffer from residual urine" and a customer-owned item of "sex, age, unmarried" (to be exact, a supplier-owned item). Send to Company X. Here, the customer-owned items are such as “male/female in sex, ages in teens/20s/30s/40s/50s/60s, unmarried/married (including separation and bereavement)” The information is not personal information such as “Customer A of Company Y is male, 30s, unmarried”.

Then, by inputting these items into the customer estimation device 1 on the company X side, not only the prospective customers regarding the indexed items out of all the customers but also the possibility at present is low, but the demand in the future Estimate potential customers (targets) that are estimated to be present.

The indexing item is acquired by the indexing item acquiring unit 2 of the customer estimation device 1 (corresponding to “indexing item acquiring means” of the present invention). Further, the acquisition of the customer possession item is performed by the customer possession item acquisition unit 3 (corresponding to “customer possession item acquisition means” of the present invention) of the customer estimation device 1.

Then, when the indexing item and the customer possession item are input to the customer estimation device 1, the item correspondence unit 4 (corresponding to the “item correspondence means” of the present invention) of the customer estimation device 1 is an area where the entire market can be estimated. Items are matched with single source data (hereinafter referred to as source data) that employs random sampling. Further, the customer estimation device 1 classifies the customer possessed items and gives a score to each classified item.

The classification of customer possessed items is performed by the classification item creation unit 5 of the customer estimation device 1 (corresponding to “classification item creation means” of the present invention). The score is assigned by the score assigning unit 6 of the customer estimation device 1 (corresponding to “score assigning means” of the present invention). The classification item creating unit 5 and the score assigning unit 6 are machine learning models obtained by machine learning using the source data as input data. Finally, the company X sends the list 15 including the classification items and the scores to the company Y. This can be said to be the provision of information on indexed items rather than information on potential customers.

Supplementing the source data, this is information that comprehensively investigates the consciousness, attributes, product involvement, media contact, etc. of a specific or unspecified consumer by a questionnaire. That is, the source data is the information that strongly captures the behavior pattern of the consumer, which is so-called, capturing the consumer from the three viewpoints of “person”, “thing”, and “media”.

In addition, the above-mentioned "consciousness" means a consumer's sense of values, tastes, consumption behavior, and the like. In addition, "attribute" is, to be exact, a demographic attribute, such as gender, age (age), area of residence, industry and annual income if working, unmarried or married, how many children, etc. Means the socio-economic characteristic data of.

Further, the above-mentioned "product involvement" means data such as clothes, electrical appliances and other brands, hobbies and special skills, and "media contact" means whether the medium used to obtain information is the Internet or a newspaper, Or, it means data such as a magazine.

In addition, this source data was extracted in the Tokyo area (within 50 km of Tokyo), Kansai area, Nagoya area, northern Kyushu area, Sapporo area, Sendai area, and Hiroshima area, which are 7 major areas in Japan, and extracted about 10,000 people. It was obtained by lending and implementing a survey-only tablet to the. The source data is not limited to questionnaire responses, and may include viewing log data (history information) such as audience ratings.

Next, FIG. 2 shows an example of the classification items (list 15) created from the customer possession items by the customer estimation device 1. Specifically, the customer estimation device 1 classifies the customer-held items by the above-described classification item creation unit 5 (rules included in the machine learning model) with respect to the “residual urine sensitivity” related to the indexed items. Go. As a result, "Gender [Male] & Age [60s]" of category item "1", "Gender [Male] & Age [40s] & "Unmarried [Unmarried]" of category item "2", etc. Individual classification items are created.

Further, the customer estimation device 1 gives a score to each classification item. This score is calculated from the numerical value of each classification item with respect to the reference value when the reference value for all the customers is calculated and set to 1 (relative evaluation). For example, the classification item “1” is a group showing a large numerical value that is about 3.8 times as large as the reference value, and a customer having a high score correlated with the indexing item can be determined as a potential customer.

For this reason, it is most efficient to take an approach such as DM distribution of a new drug to the customer of the classification item “1”. The company Y also inputs the residual urine sensitivity for each customer in the customer list L by using the list 15 including the information on the classification items and the scores.

FIG. 3 shows an example of the customer list L′ in which the items of “sensitivity of residual urine” and “determination” are newly added. As shown, when the residual urine sensitivity for each customer (the score in FIG. 2 is applied) is input, the level of the score becomes obvious. Therefore, the company Y determines that a member having a high residual urine sensitivity is “◯” (for example, a threshold value of 2.00), and DM-distributes a new drug to the person. As described above, by using the customer estimation device 1, the customer profile can be accurately estimated without obtaining personal information, cookie information, or the like that is difficult to handle.

FIG. 4 shows a result of verification as to whether or not a residual urine feeling is actually felt for a potential customer estimated by the customer estimation device 1. Specifically, the source data was divided into 4:1 and 4/5 was used for score calculation and 1/5 was used for estimation of potential customers.

The lower part of the figure shows a group with a low residual urine sensitivity calculated using the customer estimation device 1, and the higher the figure, the higher the residual urine sensitivity. As shown in the figure, the higher the figure, the higher the percentage of customers who feel that they have a feeling of residual urine. Therefore, it can be said that the indexing by the customer estimation device 1 and the estimation of potential customers have been successful.

Next, with reference to FIG. 5, a general flowchart until the client company (for example, the company Y described above) acquires the information of the potential customer by using the customer estimation device 1 will be described.

First, a service providing company (for example, the above-mentioned company X) transmits a list of source data items to the company Y (STEP 10). Through this step, the service providing company introduces a service that can estimate a potential customer by using the customer estimation device 1.

In response to this, the client company selects an indexing item from the items of the source data and sends it to the service providing company (STEP 20). The indexing items are not limited to the provision of products and services provided by client companies, but are broad themes such as whether or not there are people with the same hobbies and whether or not they correspond to people called "trend freaks". Good.

Next, the client company transmits the customer-held item (provided-party-held item) to the service providing company (STEP 30). Customer-owned items include "gender", "age (age)", "final education", "students/workers", "area of residence", etc. The possibility of matching (corresponding) increases, and the accuracy of estimation improves.

Next, the service providing company inputs the transmitted indexing items and customer possessed items into the customer estimation device 1 (STEP 40), and subsequently adjusts the customer possessed items and the source data items (STEP 50).

In STEP 50, for example, when the two items do not completely correspond to each other but there are similar items, the work of aligning the meanings is performed. For example, when the source data item includes an item related to “favorite program genre” and the customer-owned item includes “view the relevant genre three or more times a month (history information)”, both items are treated as synonymous.

Further, when adjusting both items, it may be adjusted to the item of a superordinate concept (for example, the one having a coarser granularity). For example, if the source data item includes an item "house built (new) or house owned (used)" and the item owned by the customer is house "owned", the latter can be said to be a conceptual item. Therefore, the latter item is adopted. By performing the above item adjustment work, the source data item and the customer-owned item are linked well, so it is possible to create a classification item based on the relationship between the adjusted item and the indexed item. ..

Next, since the classification rule is created by machine learning in the customer estimation device 1 (STEP 60), the classification item is created, and subsequently the score is totaled based on the rule (STEP 70). Specifically, a list (see FIG. 2) in which a score is given to each classification item is created.

After that, the service providing company transmits the list created in the above step to the client company (STEP 80).

Next, the client company gives a score to each ID (or customer number) in the customer list (STEP 90). Specifically, the client company receives the list from the service providing company, inputs the same score for each customer by the program (or by the work of the person in charge), and indexes it (see FIG. 3).

Finally, the client company gives “◯” or “x” based on the score. Here, “◯” means that the customer is a potential customer, and “◯” is given when the score exceeds (or exceeds) a predetermined threshold value. In this way, by inputting the indexing item and the customer possession item, the customer estimation device 1 can estimate the potential customer. This is the end of the series of processes for acquiring the information of the potential customer.
[Second Embodiment]

In the first embodiment, the client company (company Y) transmits the indexing item and the customer possession item to the service providing company (company X) to have the customer estimating device 1 execute the work, and receives the list 15. It was a mode. However, it is also possible to create an environment in which the customer estimation device can be used by the device terminal (PC, tablet terminal, smartphone, etc.) of the client company.

Alternatively, an employee of the client company may access a specific website and use the application of the customer estimation device 1. By doing so, the work of transmitting and receiving various items and lists is omitted, and there is no risk of information leakage. In addition, the list of source data items can be viewed on the website by the employee.

In the second embodiment described below, a client company that handles health foods (hereinafter, referred to as company Z) starts to advertise a diet supplement S (hereinafter referred to as supplement S), which is about to be sold on an EC site. Therefore, a case where there is a demand for identifying a person who is interested in it will be described as an example.

As shown in FIG. 6, the company Z has a member DB 30, and as the member information of the member list M, the gender, the date of birth, the purchase history, and favorite fruits are held. However, we do not have the essential information such as "people who are interested in dieting" and "people who want to lose weight".

In such a case, the company Z inputs the indexing item “I am interested in dieting” and the customer holding items “Gender, date of birth, purchase history, favorite fruit” to the customer estimation device 11 (application). ..

The internal block of the customer estimation device 11 is basically the same as that of the customer estimation device 1 (first embodiment), but the customer estimation device 11 relates to work that the customer estimation device 1 did not handle in the first embodiment. , And a customer determination unit 7 and a threshold value changing unit 8.

The customer determination unit 7 (corresponding to the “customer determination means” of the present invention) is not only a prospective customer regarding indexed items from all members, but it is unlikely at present but there is a demand in the future. It is possible to judge the potential customers to be sold.

Further, the threshold changing unit 8 changes the threshold of the score determined as a potential customer according to the content of the indexing item. When it is desired to narrow down the potential customers to some extent in consideration of the indexing items, the threshold changing unit 8 makes a change such as raising the threshold. This makes it possible to accurately estimate a potential customer and to advertise the supplement S.

Next, FIG. 7 shows an example of classification items (list 15') created by the customer estimation device 11 from customer possessed items. Specifically, the customer estimation device 11 classifies the customer-owned items by the classification item creation unit 5 (rules included in the machine learning model) with respect to “I am interested in diet” related to the indexed items. As a result, 20 items such as “1” for female, late 20's, have mid-year purchase, like oranges, and “2” for female, 18-22 years old, have mid-year purchase, like oranges, etc. A classification item is created.

Further, the customer estimation device 11 assigns a score to each classification item, and ranks the classification items with a high score so that they are above the list. By this work, it can be seen that the category item having a high score is a group having a high priority order for advertising and can be a potential customer.

For example, with a score of 1.50 (1.50 times the reference value) as a threshold value, classification items of 1.50 or higher are determined as potential customers. Further, a customer whose score is in the top 20% may be determined as a potential customer. For a customer determined to be a potential customer, DM distribution or the like is introduced to the supplement S, or a popup is output when the potential customer logs in to the EC site. This makes it possible to efficiently appeal the supplement S to customers who are “interested in dieting”.

This time, since the potential customer is directly determined from the score threshold, the list 15' does not include the "determination" column. However, the determination of the potential customer is performed by the customer determination unit 7 of the customer estimation device 11.

Finally, a modification of the second embodiment will be described with reference to FIGS.

FIG. 8 is a customer estimation of “gender, age, unmarried, occupation” as a customer-owned item from the member information of the member DB, and “trend freak” which is sensitive to fashion and actively transmits information as an indexed item. The member list N input to the device 11 and estimated whether or not it is a “trend freak” is shown.

Here again, the customer estimation device 11 adjusts the customer possession item and the source data item, and gives a score to each ID of the member list N. Then, when the score exceeds a predetermined threshold value (or becomes equal to or more than the threshold value), it is determined to be “trend freak (TF)” and “◯” is given.

FIG. 9 is a conceptual diagram for extracting “trend freaks” from the members. In the original state (before determination), the appearance rate of “Trend Freak (TF)” among all members was 40% (4 out of 10). After that, the customer estimation device 11 gave a score and rearranged in the order of the scores, and determined that the member whose score exceeded the threshold value (or became the threshold value or more) was “trend freak”.

At this time, for the group (5 persons) whose judgment was “O”, the actual appearance rate of “trend freak” was 60%. On the other hand, for the group (5 persons) whose judgment was “x”, the actual appearance rate of “trend freak” was 20%. As described above, by using the customer estimation device 11, the result that the efficiency increase degree is 1.5 times is obtained.

The relationship between the original appearance rate and the efficiency increase is
(Original appearance rate)×(efficiency increase rate)=(appearance rate after determination)... (Equation 1)
Given in. If the "trend freak" is to be accurately extracted, the threshold changing unit 8 may change the score threshold. On the other hand, when it is desired to increase the number of targets to be approached (the number of trend freaks extracted this time), there is a trade-off with accuracy, but the threshold value may be lowered.

In this example (modification), the score is directly calculated without making the classification items from the customer-owned items, and the “trend freak” is determined. That is, the classification item may not necessarily be created by the classification item creation unit 5.

Further, the customer estimation device 11 of the second embodiment is the same as the customer estimation device 1 of the first embodiment in that the customer profile can be accurately estimated without personal information or cookie information. is there. Rather, it is safer than the customer estimation device 1 in that it is not necessary to send and receive data such as indexed items, customer-owned items (provider-owned items), and lists.

The contents of the list, the calculation method of the score, and the communication method of the data item in the above-described embodiment are merely examples, and can be appropriately changed according to the use, purpose and the like. The communication of the first embodiment may be performed in such a manner that a person in charge receives various items sent by being attached to an e-mail and performs work. In addition, the transmitted various items may be temporarily stored on the server, and thereafter, various items may be automatically input to determination.

The item corresponding unit 4 of the customer estimation device 1 (customer estimation device 11) has been shown as an example of adjustment so as to match the item of the superordinate concept, but it is also effective to adopt the item whose freshness is newer.

In the above-described embodiment, an example is shown in which a person with a final high score is the target of the approach. However, a person with a low score also finds a value in the low, and a target of other approaches. Can be In addition, a person with a “low” score may be excluded from the approach and used for cutting off.

1, 11... Customer estimation device, 2... Indexed item acquisition unit, 3... Customer possessed item acquisition unit, 4... Item correspondence unit, 5... Classification item creation unit (machine learning model), 6... Score assigning unit, 7... Customer determination unit, 8... Threshold value changing unit, 15, 15'... List, 20... Customer DB, 30... Member DB.

Claims (5)

Indexing item acquisition means for acquiring indexing items to be indexed,
A customer-owned item acquisition means for acquiring a customer-owned item owned by the customer;
The customer-owned item acquired by the customer-owned item acquisition means and a survey conducted on a specific or unspecified person or acquired from the history information of the person, and at least the attribute, the sense of value, the taste, and the consumption of the person. Item corresponding means for making the source data item of the source data in which any one of the action and the action pattern are reflected ,
A machine learning model in which the source data items corresponding to the customer-owned items are machine-learned as input data,
Classification item creating means for classifying the customer-owned items using the machine learning model according to the contents of the indexing items and creating the classification items,
Score assigning means for assigning a score to the classification item created by the classification item creating means based on a predetermined rule using the machine learning model ,
A customer estimation device comprising: 2. The customer estimation device according to claim 1, wherein the item handling unit adjusts the items held by the customer and the source data items to be aligned with the items of the superordinate concept when the contents of the source data items conceptually overlap with each other. .. The customer estimation device according to claim 1 or 2 , wherein the score assigning unit calculates a reference value for all the customers, and calculates the score from a numerical value of the classification item for the reference value. The customer estimation device according to claim 1 or 2 , further comprising a customer determination unit that determines a potential customer when the score assigned to each of the classification items exceeds a predetermined threshold. A customer estimation method for estimating a target customer who is interested in a product or service, using a computer equipped with a machine learning model,
A data acquisition unit of the computer, an indexing item acquisition step of acquiring an indexing item to be indexed,
The data acquisition unit, a customer-owned item acquisition step of acquiring a customer-owned item owned by the customer,
The data processing unit of the computer is acquired from the customer possessed item acquired in the customer possessed item acquisition step and a survey conducted on a specific or unspecified person or history information of the person, and at least the person. An item corresponding step of making the source data item of the source data reflecting any of the attribute, the sense of value, the preference, the consumption behavior, and the behavior pattern correspond,
A machine learning model creating step of creating a machine learning model by machine learning using the source data items corresponding to the customer owned items as input data,
A classification item creating step in which the machine learning model classifies the customer-owned items according to the contents of the indexing items and creates classification items;
The machine learning model, the classification items created in the classification item creation steps, a scoring step to impart score based on a predetermined rule,
A customer estimation method comprising:

Images