Customer analysis device, customer analysis method, and program

Abstract

In a customer analysis device, a pattern-extracting means extracts transition patterns of purchase behavior of good customers on the basis of purchase histories of good customers. An analysis means generates and outputs analysis results including information on good customer candidates on the basis of customers' purchase histories and transition patterns.

Description

[Technical Field]

[0001] The present disclosure relates to analyzing customer purchasing behavior. [Background technology]

[0002] Methods have been proposed for predicting user behavior and product purchases using machine learning models. Patent Document 1 describes a method for predicting changes in customer value, which is an index of the profit a user brings to a service over a predetermined period of time. Patent Document 2 describes a method for predicting the probability that a user will enter a churn state, in which the user does not purchase any products over a predetermined period of time. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2021-189563 [Patent Document 2] Japanese Patent Publication No. 2020-177262 Summary of the Invention [Problem to be solved by the invention]

[0004] An object of the present disclosure is to provide a customer analysis device capable of analyzing a customer from the perspective of whether the customer has the potential to become a good customer. [Means for solving the problem]

[0005] In one aspect of the present disclosure, a customer analysis device includes: a rule generating means for generating rules that define conditions related to customer attributes and purchase histories and purchase behaviors performed by customers that meet the conditions, based on the purchase histories of a plurality of customers; Based on the purchase history of our loyal customers, a plurality of rules that are satisfied by the purchase history of the preferred customer; A pattern extraction means for extracting a transition pattern of purchasing behavior of good customers; Based on the degree of match of the customer's purchase history with the transition pattern an analysis means for generating and outputting an analysis result including information on potential good customers; Equipped with 、 The degree of match is the degree to which the purchase history of the customer satisfies a plurality of rules included in the transition pattern. .

[0006] In another aspect of the present disclosure, a computer-implemented method for customer analysis includes: Generate rules that define conditions related to customer attributes and purchase histories and purchase behaviors performed by customers that meet the conditions, based on the purchase histories of multiple customers; extracting a transition pattern of the purchase behavior of a good customer based on the purchase history of the good customer, the transition pattern including a plurality of rules that satisfy the purchase history of the good customer; generating and outputting an analysis result including information on potential good customers based on the degree of match of the purchase history of the customer with the transition pattern; The degree of match is the degree to which the purchase history of the customer satisfies a plurality of rules included in the transition pattern. .

[0007] In yet another aspect of the disclosure, a program includes: Generate rules that define conditions related to customer attributes and purchase histories and purchase behaviors performed by customers that meet the conditions, based on the purchase histories of multiple customers; extracting a transition pattern of the purchase behavior of a good customer based on the purchase history of the good customer, the transition pattern including a plurality of rules that satisfy the purchase history of the good customer; generating and outputting an analysis result including information about potential good customers based on the degree of match of the customer's purchase history with the transition pattern; The computer executes the process 、 The degree of match is the degree to which the purchase history of the customer satisfies a plurality of rules included in the transition pattern. . [Effects of the Invention]

[0008] According to the present disclosure, it is possible to analyze customers from the perspective of whether they have the potential to become good customers. [Brief explanation of the drawings]

[0009] [Figure 1] 1 is a block diagram showing the overall configuration of a customer analysis system according to a first embodiment. [Figure 2] FIG. 2 is a block diagram showing a hardware configuration of the customer analysis device. [Figure 3] 1 is a block diagram showing a functional configuration of a customer analysis device according to a first embodiment. [Figure 4] An example of a purchase history by customer is shown below. [Figure 5] An example of a rule is shown below. [Figure 6] An example of a preferred customer route is shown below. [Figure 7] 10 shows an example of display of the analysis results by the analysis unit. [Figure 8] 10 is a flowchart of a customer analysis process. [Figure 9]FIG. 10 is a block diagram showing the functional configuration of a customer analysis device according to a second embodiment. [Figure 10] 10 is a flowchart of a process performed by a customer analysis device according to a second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, preferred embodiments of the present disclosure will be described with reference to the drawings. First Embodiment [Overall configuration] 1 is a block diagram showing the overall configuration of a customer analysis system according to a first embodiment. The customer analysis system 1 includes a customer analysis device 100, a customer database (hereinafter, "database" will be abbreviated as "DB") 2, and a purchase DB 3. The customer analysis device 100 is configured by a computer such as a personal computer (PC).

[0011] The customer DB2 stores customer information about customers who visit the store. Specifically, the customer information includes attributes of each customer associated with a customer ID. The customer attributes include, for example, age, gender, occupation, family structure, place of residence, housing type, and workplace information. The customer information also includes classification information indicating whether each customer is a preferred customer or a general customer other than a preferred customer.

[0012] The purchase DB3 stores purchase information of products by customers who visit the store. Specifically, the purchase information includes the purchase date and time, customer ID, name, price, and category (type) of the purchased product for each customer at the store. For example, POS (Point of Sales) data for each store can be used as the purchase information.

[0013] The customer analysis device 100 analyzes the purchase history of good customers based on the customer information stored in the customer DB 2 and the purchase information stored in the purchase DB 3, and extracts transition patterns of the purchase behavior of good customers.The customer analysis device 100 then compares the transition patterns of the customer's purchase behavior with the transition patterns of the purchase behavior of good customers to analyze whether the customer has the potential to become a good customer, and outputs the analysis results.

[0014] [Hardware configuration] 2 is a block diagram showing the hardware configuration of the customer analysis device 100. As shown in the figure, the customer analysis device 100 includes an interface (I / F) 11, a processor 12, a memory 13, a recording medium 14, a database (DB) 15, a display unit 16, and an input unit 17.

[0015] The I / F 11 inputs and outputs data to and from external devices. Specifically, customer information stored in the customer DB 2 and purchase information stored in the purchase DB 3 are input to the customer analysis device 100 via the I / F 11.

[0016] The processor 12 is a computer such as a CPU (Central Processing Unit) and controls the entire customer analysis device 100 by executing a pre-prepared program. The processor 12 may be a GPU (Graphics Processing Unit) or an FPGA (Field-Programmable Gate Array). The processor 12 executes the customer analysis process described below.

[0017] The memory 13 is configured by a ROM (Read Only Memory), a RAM (Random Access Memory), etc. The memory 13 is also used as a working memory while the processor 12 is executing various processes.

[0018] The recording medium 14 is a non-volatile, non-temporary recording medium such as a disk-shaped recording medium or semiconductor memory, and is configured to be detachable from the customer analysis device 100. The recording medium 14 records various programs executed by the processor 12. When the customer analysis device 100 executes various processes, the programs recorded on the recording medium 14 are loaded into the memory 13 and executed by the processor 12.

[0019] DB15 stores data generated by the customer analysis device 100. Specifically, DB15 stores customer-specific purchase history, rules, and good customer routes, which will be described later. In addition, when the customer analysis device 100 uses a machine learning model, information on the trained model is stored in DB15.

[0020] The display unit 16 is, for example, a liquid crystal display device, and displays the prediction information generated by the customer analysis device 100. The input unit 17 is, for example, a mouse, a keyboard, and the like, and is used by the user to give instructions and input necessary when viewing the prediction information of the model.

[0021] (Functional configuration) 3 is a block diagram showing the functional configuration of the customer analysis device 100 according to the first embodiment. Functionally, the customer analysis device 100 includes a customer-specific purchase history generation unit 21, a rule generation unit 22, a preferred customer route generation unit 23, and an analysis unit 24.

[0022] The customer-specific purchase history generation unit 21 generates a customer-specific purchase history based on the customer information stored in the customer DB 2 and the purchase information stored in the purchase DB 3. Figure 4 shows an example of a customer-specific purchase history. The customer-specific purchase history is information that collects product purchase histories for each customer, and includes customer attributes associated with the customer ID, purchase history, and customer classification.

[0023] "Attributes" refers to information such as age and gender obtained from the customer database 2. "Purchase history" refers to each customer's purchase history obtained from the purchase database 3, including the date and time of each purchase and information on the products purchased. "Customer classification" refers to the classification of customers from the store's perspective, specifically, whether each customer is a preferred customer or a general customer. For example, a preferred customer may be a customer who has made purchases of more than a specified amount at the store within a specified period, or a customer who visits the store more frequently than a specified frequency. Alternatively, a preferred customer may be a customer who exhibits significant purchasing behavior, such as a customer who frequently shops outside of sales periods or a customer who purchases a specific product at a high price or with a high purchase frequency. On the other hand, a general customer is any customer other than a preferred customer. Customer classification is not limited to two categories: preferred customers and general customers. For example, a three- or five-level customer ranking may be used based on purchase history. The customer-specific purchase history generation unit 21 stores the generated customer-specific purchase history in the database 15 and outputs it to the preferred customer route generation unit 23 and the analysis unit 24.

[0024] The rule generation unit 22 generates rules regarding product purchases by customers based on the purchase information stored in the purchase DB 3. Here, "rules" refer to rules and tendencies in customer purchasing behavior. Rules indicate relationships such as, for example, that customers with certain attributes and purchase histories tend to purchase certain products or tend to spend more on them.

[0025] FIG. 5 shows an example of a rule. In this embodiment, an IF-THEN rule is used as the rule. As shown in the figure, each rule includes a rule number, an application condition, a result, and the number of cases (number of people) that satisfy the rule. The "rule number" is an identification number uniquely assigned to each rule. The "application condition" is a condition for applying the rule, and corresponds to the IF in the IF-THEN rule. The application condition is a condition related to at least one of the customer's attributes and purchase history. Regarding the purchase history, the condition includes the category of purchased products, the number of purchases of products, the average purchase interval, etc. The "result" is the purchasing behavior that occurs for a customer when the rule is satisfied, and corresponds to the THEN in the IF-THEN rule. The result is information such as purchasing products in a specific category or the amount of increase in purchase amount. The "number of cases" is the number of customers that have satisfied each rule in the past.

[0026] The rule generation unit 22 analyzes the purchasing information acquired from the purchasing DB 3, extracts cases in which common results occur under common application conditions, and generates rules that specify the application conditions and results when a predetermined number or a predetermined percentage of such cases exist. The rule generation unit 22 may generate rules using a machine learning model. For example, the rule generation unit 22 may extract rules from the purchasing information using a trained model that has learned the relationship between application conditions related to customer attributes and purchase history and purchasing behavior that occurs when the application conditions are met. By applying the extracted rules to the customer's purchase history, it is possible to predict what purchasing behavior the customer will exhibit in the future. The rule generation unit 22 stores the extracted rules in the DB 15 and outputs them to the preferred customer route generation unit 23 and the analysis unit 24.

[0027] The preferred customer route generation unit 23 generates a preferred customer route based on the customer purchase history and the rules. The preferred customer route indicates the route of purchase behavior that a general customer has taken to become a preferred customer, i.e., the transition pattern of multiple rules.

[0028] Figure 6 shows an example of a preferred customer route. A preferred customer route includes a route number and a transition pattern. The "route number" is an identification number that uniquely identifies a preferred customer route. The "transition pattern" indicates the transition pattern of rules that a preferred customer has satisfied in the past. For example, preferred customer route 01 indicates that many general customers became preferred customers as a result of satisfying rules R1, R5, and R10 in that order. Furthermore, preferred customer route 02 indicates that many general customers became preferred customers as a result of satisfying rules R2, R11, and R7 in that order. It can be predicted that general customers who have engaged in purchasing behavior that matches any of the preferred customer routes are likely to become preferred customers.

[0029] The good customer route generation unit 23 generates a good customer route based on the purchase history of good customers included in the customer-specific purchase history and the rules generated by the rule generation unit 22. Specifically, the good customer route generation unit 23 extracts a transition pattern of rules that a predetermined number or a predetermined percentage or more of good customers have commonly satisfied in the past, i.e., a combination and order of multiple rules, and generates a good customer route. The good customer route generation unit 23 may generate a good customer route using a machine learning model. Specifically, the good customer route generation unit 23 may generate a good customer route using a trained model that predicts a good customer route based on each customer's purchase history, rules, and information on whether each customer is a good customer. The good customer route generation unit 23 outputs the generated good customer route to the analysis unit 24. The good customer route generation unit 23 is an example of a pattern extraction unit.

[0030] The analysis unit 24 analyzes the purchasing behavior of customers using the purchase history by customer, the rules, and the good customer route, and predicts customers who are likely to become good customers (hereinafter also referred to as "good customer candidates"). Specifically, the analysis unit 24 extracts, from among the general customers included in the purchase history by customer, general customers whose purchase history matches the good customer route up to a certain point, as good customer candidates.

[0031] The analysis unit 24 references the purchase history of each customer and determines whether the purchase history of each general customer satisfies each rule included in the preferred customer route, and determines the degree of match with the preferred customer route. The analysis unit 24 then predicts general customers whose degree of match with the preferred customer route is at or above a predetermined level as preferred customer candidates. The analysis unit 24 then analyzes whether each general customer is a preferred customer candidate for each preferred customer route, and outputs the analysis results to the display unit 16. At this time, the analysis unit 24 can simply list the names and attribute information of the predicted preferred customer candidates and output them. The analysis unit 24 can also output detailed analysis results, as described below. The display unit 16 displays the input analysis results. This allows store promotion staff and others to obtain information on customers who are likely to become preferred customers. The analysis unit 24 is an example of an analysis means.

[0032] [Example of analysis results] FIG. 7 shows an example of the display of the analysis results by the analysis unit 24. This display example includes a good customer route 30 and analysis information 40. The analysis unit 24 makes predictions for each good customer route, and the good customer route 30 indicates the good customer route used to obtain the analysis results. That is, in the example of FIG. 7, the analysis results using the good customer route with route number 02 are displayed.

[0033] The preferred customer route 30 includes a route number 31, a transition pattern 32, and a number of people 33. The route number 31 is a number that indicates the preferred customer route. The transition pattern indicates the transition pattern of the rules in the preferred customer route. In the example of Figure 7, the preferred customer route 30 with route number 02 is a route that satisfies rules R2, R11, and R7 in this order. The number of people 33 indicates the number of customers that were the subject of the analysis.

[0034] The analysis information 40 includes a level 41, a segment 42 of potential good customers, extraction conditions 43, triggers 44 to become good customers, number of people 45, number of purchasers 45, average increase value 47, and reference increase value 48. The level 41 indicates the level of the potential good customers, i.e., how close they are to becoming good customers. Note that the higher the level number, the closer they are to becoming good customers.

[0035] The prime customer candidate segment 42 indicates a segment (group) of customers classified as prime customer candidates. In the example of Figure 7, three segments are set based on the degree of match with the prime customer rules. The prime customer route 30 is a route that satisfies rules R2, R11, and R7, in that order. The level 0 segment includes general customers who do not satisfy the first rule R2. The level 1 segment includes general customers who satisfy rules R1 through R2. The level 2 segment includes general customers who satisfy rules R1 through R11. In this way, the analysis results can present prime customer candidates categorized into different levels depending on their progress along the prime customer route, i.e., how far along the prime customer route they have satisfied. Note that general customers belonging to the level 0 segment do not qualify as prime customer candidates because their purchasing behavior does not match the prime customer route. On the other hand, general customers belonging to the level 1 and level 2 segments qualify as prime customer candidates because some of their purchasing behavior matches the prime customer route.

[0036] The extraction conditions 43 indicate the conditions for extracting general customers belonging to each level of segment. These conditions are set using, for example, the number of purchases of a specific category or the average purchase interval. The trigger 44 for becoming a good customer indicates the next purchase behavior that general customers belonging to each segment need to take to become good customers. In other words, the trigger 44 for becoming a good customer indicates the purchase behavior proposed to guide potential good customers to become good customers. The trigger 44 for becoming a good customer is determined based on the good customer route. For example, since a general customer belonging to a level 0 segment has not yet satisfied rule R2, a purchase behavior to satisfy rule R2, i.e., purchasing category B, is presented as a trigger for becoming a good customer. Similarly, a general customer belonging to a level 1 segment has satisfied rule R2 but has not yet satisfied rule R11, so a purchase behavior to satisfy rule R11, i.e., purchasing category F, is presented as a trigger for becoming a good customer.

[0037] The number of people 45 indicates the number of general customers belonging to each segment. In the example of Figure 7, there are 705 customers belonging to segment 0, 538 customers belonging to segment 1, and 238 customers belonging to segment 2. The total number of customers belonging to each of segments 0 to 2 is shown in the number of people 33 of the preferred customer route 30.

[0038] The number of purchasers 45 indicates the number of people who purchased a product in accordance with the purchasing behavior indicated in the trigger 44 for good customers. When the trigger 44 for good customers is presented, the store implements sales measures to encourage that purchasing behavior. For example, the store may offer a discount on category B to customers belonging to the level 0 segment. The number of purchasers 45 is the number of customers who performed the purchasing behavior indicated in the trigger 44 for good customers in response to such measures. The average increase 47 indicates the average increase in purchase amount by customers who performed purchasing behavior in response to the above measures. In this example, the purchase of category B was presented as the trigger 44 for good customers, and the store offered a discount on category B to customers belonging to the level 0 segment. As a result, 283 people purchased category B, and the average purchase amount of these customers increased by 1,666.3 yen.

[0039] Reference increase value 48 indicates the average increase in purchase amount of potential good customers based on past analysis. In other words, when a past analysis of potential good customers was conducted for the good customer route with route number 02, the average increase in purchase amount of customers belonging to segments 0 to 2 was 950.5 yen, 1778.5 yen, and 1322.4 yen, respectively.

[0040] [Customer analysis processing] Next, the customer analysis process executed by the customer analysis device 100 will be described. Fig. 8 is a flowchart of the customer analysis process. This process is realized by the processor 12 shown in Fig. 2 executing a pre-prepared program and mainly operating as the analysis unit 24 shown in Fig. 3. It is assumed that the purchase history by customer and the good customer route have already been generated.

[0041] First, the analysis unit 24 selects one good customer route from multiple good customer routes (step S10). In the example of FIG. 7, the analysis unit 24 selects the good customer route with route number 02. Next, the analysis unit 24 acquires multiple rules included in the selected good customer route (step S11). In the example of FIG. 7, the analysis unit 24 acquires rules R2, R11, and R7 included in the good customer route with route number 02.

[0042] Next, the analysis unit 24 references the purchase history for each customer and extracts customers who satisfy each rule from among the customers who meet the extraction conditions (step S12). Next, the analysis unit 24 classifies the customers based on their degree of conformance to the good customer route, counts the number of customers, and generates analysis results (step S13). Specifically, as shown in FIG. 7, the analysis unit 24 classifies potential good customers into multiple levels of segments based on their degree of conformance to the good customer route, i.e., the rules they satisfy. Then, for each customer belonging to each segment, the analysis unit 24 calculates the trigger 44 to become a good customer, the number of customers 45 and 46, and the increase values ​​47 and 48, as illustrated in FIG. 7, and generates analysis results. Then, the analysis unit 24 outputs the analysis results to the display unit 16 for display (step S14). In this way, the customer analysis process for one good customer route is completed. Note that the analysis unit 24 can select another good customer route and perform a similar analysis.

[0043] [Variations] (Variation 1) In the example of Figure 7, the analysis unit 24 determines that customers who satisfy the conditions up to the second rule R11 are potential good customers of level 2, but it may also determine that customers who satisfy the application conditions up to the third rule R7 are potential good customers of level 3, who have an even higher probability of becoming good customers.

[0044] (Variation 2) In the above embodiment, the analysis unit 24 determines the level of a potential good customer based on the degree of match with the good customer route, i.e., the degree to which the target customer satisfies the multiple rules included in the good customer route in the specified order. Alternatively, the level of a potential good customer may be determined based on how many of the multiple rules the target customer satisfies, regardless of the order in which the rules are satisfied. For example, in the case of the good customer route with route number 02 shown in FIG. 7, the analysis unit 24 may determine a customer who does not satisfy any of rules R2, R11, and R7 as level 0, a customer who satisfies one of the rules as level 1, and a customer who satisfies two of the rules as level 2.

[0045] (Variation 3) In the above embodiment, the preferred customer route is described as including three rules, but this is not limited thereto, and the preferred customer route may include any number of rules. Also, multiple preferred customer routes may include different numbers of rules. For example, one preferred customer route may include three rules, and another preferred customer route may include four rules.

[0046] Second Embodiment 9 is a block diagram showing the functional configuration of a customer analysis device 70 according to the second embodiment. The customer analysis device 70 includes a pattern extraction unit 71 and an analysis unit 72.

[0047] 10 is a flowchart of processing by the customer analysis device 70 of the second embodiment. The pattern extraction means 71 extracts transition patterns of purchase behavior of good customers based on the purchase history of the good customers (step S71). The analysis means 72 generates and outputs analysis results including information on good customer candidates based on the customer purchase history and the transition patterns (step S72).

[0048] According to the model analysis device 70 of the second embodiment, it is possible to analyze customers from the perspective of whether they have the potential to become good customers.

[0049] A part or all of the above-described embodiments can be described as, but not limited to, the following supplementary notes.

[0050] (Appendix 1) A pattern extraction means for extracting a transition pattern of purchase behavior of a good customer based on the purchase history of the good customer; an analysis means for generating and outputting an analysis result including information on potential good customers based on the purchase history of the customer and the transition pattern; A customer analysis device comprising:

[0051] (Appendix 2) 2. The customer analysis device according to claim 1, wherein the analysis means generates the analysis result based on a degree of coincidence of the purchase history of the customer with the transition pattern.

[0052] (Appendix 3) The pattern extraction means generates rules that define conditions related to customer attributes and purchase histories and purchasing behaviors performed by customers that meet the conditions, based on the purchase histories of a plurality of customers; the transition pattern includes a plurality of rules that are satisfied by the purchase history of the good customers; 3. The customer analysis device according to claim 2, wherein the degree of match is the degree to which the purchase history of the customer satisfies the plurality of rules.

[0053] (Appendix 4) the transition pattern includes a plurality of rules satisfied by the purchase history of the good customer and an order in which the plurality of rules are satisfied; 4. The customer analysis device according to claim 3, wherein the degree of match is the degree to which the purchase history of the customer corresponds to the plurality of rules in the order.

[0054] (Appendix 5) A customer analysis device according to any one of appendices 1 to 4, wherein the analysis results indicate information about the potential good customers for each level set based on the degree of similarity of the customer's purchase history.

[0055] (Appendix 6) The customer analysis device according to claim 5, wherein the information about the good customer candidates includes the degree of match of the good customer candidates at each level with the transition pattern and the number of good customer candidates at each level.

[0056] (Appendix 7) The customer analysis device according to any one of appendices 1 to 6, wherein the analysis results include purchasing behaviors suggested to guide potential good customers to become good customers based on the transition patterns.

[0057] (Appendix 8) The customer analysis device according to any one of appendices 1 to 7, wherein the pattern extraction means extracts the transition patterns using a trained machine learning model that predicts the transition patterns based on the attributes and purchase history of the customers and whether each of the customers is a good customer.

[0058] (Appendix 9) Extract transition patterns of purchasing behavior of good customers based on their purchase history, A customer analysis method for generating and outputting analysis results including information on potential good customers based on the customer's purchase history and the transition pattern.

[0059] (Appendix 10) Extract transition patterns of purchasing behavior of good customers based on their purchase history, A recording medium storing a program for causing a computer to execute a process of generating and outputting analysis results including information about potential good customers based on the customer's purchase history and the transition pattern.

[0060] Although the present disclosure has been described above with reference to the embodiments and examples, the present disclosure is not limited to the above-described embodiments and examples. Various modifications that can be understood by a person skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure. [Explanation of symbols]

[0061] 1. Customer analysis system 2 Customer DB 3 Purchasing DB 12 processors 16 Display section 21 Customer purchase history generation unit 22 Rule Generation Unit 23 Preferred Customer Route Generation Department 24 Analysis Department

Claims

1. A rule generation means for generating rules that specify conditions related to customer attributes and purchase histories and purchase behaviors of customers that meet the conditions, based on the purchase histories of multiple customers; A pattern extraction means for extracting a transition pattern of purchase behavior of a good customer, the transition pattern including a plurality of rules satisfied by the purchase history of the good customer, based on the purchase history of the good customer; an analysis means for generating and outputting an analysis result including information on potential good customers based on the degree of coincidence of the purchase history of the customer with the transition pattern; Equipped with The degree of match is the degree to which the purchase history of the customer satisfies a plurality of rules included in the transition pattern.

2. the transition pattern includes an order in which the purchase history of the good customer satisfies the plurality of rules; The customer analysis device according to claim 1 , wherein the degree of match is the degree to which the purchase history of the customer corresponds to the plurality of rules in the order.

3. The customer analysis device according to claim 1 or 2, wherein the analysis result indicates information about the potential good customers for each level set based on the degree of coincidence of the purchase history of the customers.

4. The customer analysis device according to claim 3 , wherein the information about the good customer candidates includes a degree of coincidence of the good customer candidates at each level with the transition pattern and the number of good customer candidates at each level.

5. The customer analysis device according to claim 1 , wherein the analysis result includes a purchasing behavior suggested for guiding a potential good customer to become a good customer based on the transition pattern.

6. 6. The customer analysis device according to claim 1, wherein the pattern extraction means extracts the transition pattern using a trained machine learning model that predicts the transition pattern based on the attributes and purchase history of the customers and whether each of the customers is a good customer.

7. 1. A computer-implemented method for customer analysis, comprising: Generate rules that define conditions related to customer attributes and purchase histories and purchase behaviors performed by customers that meet the conditions based on the purchase histories of multiple customers; extracting a transition pattern of the purchase behavior of a good customer based on the purchase history of the good customer, the transition pattern including a plurality of rules that satisfy the purchase history of the good customer; generating and outputting an analysis result including information on potential good customers based on the degree of match of the purchase history of the customer with the transition pattern; A customer analysis method in which the degree of match is the degree to which the purchase history of the customer satisfies a plurality of rules included in the transition pattern.

8. Based on the purchase histories of multiple customers, rules are generated that specify conditions regarding customer attributes and purchase histories and the purchase behavior of customers that meet the conditions; extracting a transition pattern of the purchase behavior of a good customer based on the purchase history of the good customer, the transition pattern including a plurality of rules that satisfy the purchase history of the good customer; causing a computer to execute a process of generating and outputting an analysis result including information on potential good customers based on a degree of coincidence of the purchase history of the customer with the transition pattern; The degree of match is the degree to which the purchase history of the customer satisfies a plurality of rules included in the transition pattern.

Images