JP2015146145A - Customer analyzing program, customer analyzing method and customer analyzer - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a customer analyzing program, a customer analyzing method and a customer analyzer capable of analyzing customers while considering trend of the time.SOLUTION: The customer analyzing program is a program to cause a computer to execute: to receive an instruction to set a specified period as a target period of analysis or an instruction to set a predetermined period from a present or past point as the target period of analysis, and a purchase history data of target stores of the analysis; to compare an integrated customer group including all customer groups which have predetermined similar relationship from the customer groups generated based on plural purchase history data stored in the target period of the instructed analysis and a customer group which is generated based on the purchase history data from the stores as the target of the analysis: and to output the comparison result.

Description

  This case relates to a customer analysis program, a customer analysis method, and a customer analysis apparatus.

  For example, department stores and mass retailers (hereinafter referred to as “sales stores”) analyze customers who visit the stores, and use the analysis results in management strategies such as product measures and sales floor measures. Customer analysis, for example, categorizes customers into multiple groups based on age, gender, address, etc., or categorizes them into multiple groups based on customer purchase information (product brand, price, frequency, etc.) Is done.

  Regarding the analysis means, for example, Patent Literature 1 describes a point of analyzing a person's behavior pattern. Patent Document 2 describes that selectable items are grouped.

JP2013-131170A JP 2011-28534 A

  However, customer group information obtained by customer analysis may be used for comparative analysis with general public information such as demographic and commercial statistics of the region where the store is located. , It is not used for comparative analysis with the analysis results of a plurality of other dealers. Therefore, detailed comparison analysis of customer groups using customer analysis results of other dealers is desired.

  Further, it is desirable that the comparative analysis is performed in consideration of the trend of the time with respect to the purchase contents of the customer. Note that this problem is not limited to a store that sells merchandise, and similarly exists for a store that provides a specific service, such as a restaurant.

  In one aspect, an object of the present invention is to provide a customer analysis program, a customer analysis method, and a customer analysis apparatus that can perform customer analysis in consideration of a trend of time.

  In one aspect, the customer analysis program includes an instruction to designate a period as an analysis target period, or an instruction to set a predetermined period from a current or past time point as an analysis target period, and purchase history data of a store to be analyzed. An integrated customer group including all customer groups having a predetermined similarity among customer groups generated based on purchase history data of a plurality of stores in the instructed analysis target period, and the analysis target A computer executes a process of outputting a comparison result with a customer group generated based on store purchase history data.

  In another aspect, the customer analysis program includes an instruction to designate a period as an analysis target period, or an instruction to set a predetermined period from a current or past time point as an analysis target period, purchase history data of an analysis target store, Each customer group included in an integrated customer group including all customer groups having a predetermined similar relationship among customer groups generated based on purchase history data of a plurality of stores in the analysis target period , Referring to the storage unit stored in association with the time indicated by the purchase history data based on each customer group, among the integrated customer group of the time included in the specified analysis target period Extracting an integrated customer group based on purchase history data, and based on the extracted integrated customer group and purchase history data of the analysis target store And outputs the result of comparison between the generated customer group, the process to be executed by a computer.

In one aspect, the customer analysis method includes an instruction to designate a period as an analysis target period, or an instruction to set a predetermined period from a current or past time point as an analysis target period, and purchase history data of a store to be analyzed. A process of accepting,
An integrated customer group including all customer groups having a predetermined similar relationship among customer groups generated based on purchase history data of a plurality of stores in the specified analysis target period, and purchase of the analysis target store The computer executes a process of outputting a comparison result with the customer group generated based on the history data.

  In another aspect, the customer analysis method includes an instruction to designate a period as an analysis target period, or an instruction to set a predetermined period from a current or past time point as an analysis target period, and purchase history data of a store to be analyzed. And a customer group generated based on purchase history data of a plurality of stores in the instructed analysis target period, and included in an integrated customer group including all customer groups having a predetermined similarity relationship Referring to the storage unit that stores the customer group in association with the time indicated by the purchase history data based on each customer group, the time included in the specified analysis target period of the integrated customer group A process of extracting an integrated customer group based on the purchase history data, the extracted integrated customer group, and purchase history data of the analysis target store And a step of outputting the result of comparison between the generated customer groups based, computer executes.

  In one aspect, the customer analysis device includes an instruction to designate a period as an analysis target period, or an instruction to set a predetermined period from a current or past time point as an analysis target period, and purchase history data of a store to be analyzed. An integrated customer group including all customer groups having a predetermined similarity among customer groups generated based on purchase history data of a plurality of stores in the instructed analysis target period, An output processing unit that outputs a comparison result with the customer group generated based on the purchase history data of the analysis target store.

  In another aspect, the customer analysis device includes an instruction to designate a period as an analysis target period, or an instruction to set a predetermined period from a current or past time point as an analysis target period, and purchase history data of a store to be analyzed. Included in an integrated customer group that includes a reception processing unit that accepts and all customer groups having a predetermined similar relationship among customer groups generated based on purchase history data of a plurality of stores in the instructed analysis target period Each of the customer groups to be stored in association with the time indicated by the purchase history data based on each customer group, and the storage unit, with reference to the storage unit, the specified customer of the integrated customer group The integrated customer group based on the purchase history data in the period included in the analysis target period is extracted, the extracted integrated customer group, and the purchase of the analysis target store. And an output processing unit that outputs a result of comparison between customer groups that are generated based on historical data.

  Customers can be analyzed in consideration of seasonal trends.

It is a lineblock diagram showing an example of a customer analysis system. It is a block diagram which shows an example of a terminal device. It is a table | surface which shows an example of purchase history data and customer data. It is a block diagram which shows a customer analyzer. It is a block diagram which shows an example of a function structure of the customer analyzer which concerns on 1st Example. It is a table | surface which shows the example of a universal cluster database and a private company cluster database. It is a flowchart of the customer analysis program which concerns on 1st Example. It is a figure which shows an example of the data for cluster generation, and the three-dimensional coordinate for cluster generation. It is a flowchart which shows the production | generation process of a comparison object cluster. It is a figure which shows the example of a production | generation of a comparison object cluster. It is a table | surface which shows the example of the parameter | index used for a similarity determination process. It is a graph which shows the change characteristic of a parameter | index. It is a table | surface which shows the example of determination of the similarity of a cluster. It is a table | surface which shows an example of a comparison object cluster. It is a flowchart of the comparative analysis process of the cluster in 1st Example. It is a flowchart of the construction process of a universal cluster. It is a figure which shows the example of an output of the analysis result obtained by a comparison analysis process. It is a table | surface which shows the output example of the difference of a private company cluster and a comparison object cluster. It is a figure which shows the cluster before a customer analysis process in the customer analysis example 1. FIG. It is a figure which shows the individual company cluster of a department store of an analysis object, and a comparison object cluster. It is a figure which shows the individual company cluster and comparison object cluster which have similarity. It is a block diagram which shows an example of a function structure of the customer analyzer which concerns on 2nd Example. It is a flowchart of the customer analysis program which concerns on 2nd Example. It is a figure which shows the example of extraction of a universal cluster. It is a flowchart of the comparison analysis process of the cluster in 2nd Example. It is a figure which shows the result of the comparison analysis process of the cluster in the customer analysis example 2. FIG. It is a figure which shows the private company cluster and universal cluster which have similarity in the customer analysis example 2. It is a figure which shows the change process of the universal cluster in a customer analysis example 2, and a universal cluster candidate. It is a figure which shows the cluster after the customer analysis process in the customer analysis example 2. FIG.

  FIG. 1 is a configuration diagram illustrating an example of a customer analysis system. The customer analysis system includes a customer analysis device 1 and a terminal device 2 installed in each of the department stores A to X. The terminal device 2 is a computer device such as a personal computer, for example, and is connected to the customer analysis device 1 via a network NW such as the Internet. Note that the customer analysis device 1 can execute customer analysis processing using stored data even when the customer analysis device 1 is not connected to the network NW (offline state).

  The customer analysis device 1 is a computer device such as a server device, for example, and provides a customer analysis service. The customer analysis device 1 collects various data from the terminal device 2 to analyze the customers of the department stores A to X, and transmits the analysis results of the customers of the department stores A to X to the terminal device 2.

  More specifically, the customer analysis device 1 includes an integrated customer group to which a similar customer group corresponding to a specified period among customer groups based on purchase history data of a plurality of department stores, and a customer group of a department store X to be analyzed. And the comparison result is transmitted to the terminal device 2 of the department store X. Thereby, the customer analysis apparatus 1 performs customer analysis in consideration of the trend of the time.

  In the present embodiment, the customer analysis apparatus 1 analyzes the department store X and analyzes the customers of the department store X using a database constructed based on the analysis results of other department stores A, B, C,. To do. The analysis target of the customer analysis device 1 is not limited to the department store X, and a specific service is provided to the customer such as a mass sales store, a convenience store, and another type of sales store such as a car dealer, for example, a restaurant. It may be a store that provides it.

  FIG. 2 is a configuration diagram illustrating an example of the terminal device 2. The terminal device 2 includes a CPU (Central Processing Unit) 20, a ROM (Read Only Memory) 21, a RAM (Random Access Memory) 22, an HDD (Hard Disk Drive) 23, and a communication processing unit 24. The terminal device 2 further includes a portable storage medium drive 25, an input processing unit 26, an image processing unit 27, and the like.

  The CPU 20 is arithmetic processing means and operates according to a communication processing program (software) for transmitting / receiving data to / from the customer analysis device 1. The CPU 20 is connected to the units 21 to 27 via the bus 28. The terminal device 2 is not limited to one that operates by software, and hardware such as an application specific integrated circuit may be used instead of the CPU 20.

  The RAM 22 is used as a working memory for the CPU 20. The ROM 21 and HDD 23 are used as storage means for storing programs and the like. The HDD 23 further stores purchase history data 230 indicating purchase histories of customers in the department stores A to X (customers registered as members of the department stores) and customer data 231 indicating attributes of the customers of the department stores A to X.

  FIG. 3 shows an example of purchase history data 230 (see FIG. 3A) and customer data 231 (see FIG. 3B). The purchase history data 230 includes “store name” information, “department” information, “brand name” information, “purchase date” information, “purchase time” information, and “customer ID”.

  “Store name” information is identification information of department stores A to X. The “store name” information is common to each column for each of the department stores A to X, and is used by the customer analyzer 1 to distinguish the department stores A to X.

  The “department” information indicates the department of the product for which purchase has been performed. The “brand name” information indicates the brand name (brand A, B, etc.) of the purchased product. The “purchase date” information and the “purchase time” information respectively indicate the date and time when the purchase was performed. “Customer ID” is a customer identification number, for example, a customer membership number. “Brand name” information is only an example. Instead, “product name” information indicating the name of the purchased product, “floor name” information indicating the name of the floor where the product was purchased, or other than the product. “Service name” information indicating the service name may be used.

  The customer data 231 includes “customer ID”, “age” information, “sex” information, and “zip code”. “Customer ID” corresponds to the customer ID described above. The “age” information indicates the age of the customer. The “age” information may be shown as a certain age group (for example, teens, 20s,...) Instead of a specific age as shown in FIG. .

  “Gender” information indicates the gender of the customer. “Postal code” indicates the postal code of the customer's residence area. Instead of “zip code”, the address of the customer or a part of the address may be used.

  The CPU 20 reads the purchase history data 230 and the customer data 231 from the HDD 23 and outputs them to the communication processing unit 24. The communication processing unit 24 is, for example, a network card, and is a communication unit that communicates with the customer analysis device 1 and the like via the network NW.

  The portable storage medium drive 25 is a device that writes information to and reads information from the portable storage medium 250. Examples of the portable storage medium 250 include a USB memory (USB: Universal Serial Bus), a CD-R (Compact Disc Recordable), and a memory card. Note that the communication processing program may be stored in the portable storage medium 250.

  The terminal device 2 further includes an input device 260 for performing an information input operation and a display 270 for displaying an image. The input device 260 is input means such as a keyboard and a mouse, and input information is output to the CPU 20 via the input processing unit 26. For example, the CPU 20 receives the analysis target period designation information input by the user through the input device 260.

  The analysis target period is customer group information (comparison target cluster and universal cluster described later) used as a comparison target of customer group information (individual company cluster described later) generated based on the purchase history data 230 in the customer analysis device 1. ) Period. The analysis target period is specified (instructed) as a specific period (such as 2010) or a predetermined period (such as after 2011 or before 2009) from the current or past time point. As described above, since the analysis target period is designated from the terminal device 2, the customer analysis device 1 can perform customer analysis in consideration of the trend of the time. The analysis target period may be specified in the customer analysis device 1.

  The display 270 is image display means such as a liquid crystal display, and the displayed image data is output from the CPU 20 to the display via the image processing unit 27. In addition, it can replace with the input device 260 and the display 270, and devices, such as a touch panel provided with these functions, can also be used.

  The CPU 20 executes a program stored in the ROM 21 or the HDD 23 or a program read from the portable storage medium 250 by the portable storage medium drive 25. This program includes not only the OS (Operating System) but also the above communication processing program. The communication processing program may be downloaded via the communication processing unit 24.

  When the communication processing program is executed, the CPU 20 writes, for example, purchase history data 230 and customer data 231 input via the input processing unit 26 in the HDD 23. Note that the means for inputting the purchase history data 230 and the customer data 231 is not limited to the input processing unit 26. The purchase history data 230 and the customer data 231 may be input from other computer devices in the same department store A to X via the communication processing unit 24, or from the portable storage medium 250 to the portable storage medium drive 25. It may be input via.

  Further, for example, when receiving a request from the customer analysis apparatus 1 or when a predetermined input operation is performed with the input device 260, the CPU 20 reads the purchase history data 230 and the customer data 231 from the HDD 23 and performs customer analysis. Transmit to device 1. In addition, when an input operation for the analysis target period is performed with the input device 260, the CPU 20 transmits the analysis target period designation information to the customer analysis device 1. Further, when the CPU 20 receives data indicating the customer analysis result from the customer analysis device 1, the CPU 20 writes the data in a storage unit such as the HDD 23 and displays it on the display 270. The specification of the analysis target period is not limited to the input device 260, and may be performed by the communication processing unit 24 via a network.

  FIG. 4 is a configuration diagram showing the customer analysis device 1. The customer analysis device 1 includes a CPU 10, a ROM 11, a RAM 12, an HDD 13, a communication processing unit 14, a portable storage medium drive 15, an input processing unit 16, an image processing unit 17, and the like.

  The CPU 10 is arithmetic processing means, and executes a customer analysis method to be described later according to a customer analysis program (software). The CPU 10 is connected to the units 11 to 17 via the bus 18. The customer analysis apparatus 1 is not limited to one that operates by software, and hardware such as an application specific integrated circuit may be used instead of the CPU 10.

  The RAM 12 is used as a working memory for the CPU 10. The ROM 11 and HDD 13 are used as storage means for storing a customer analysis program for operating the CPU 10. The communication processing unit 14 is, for example, a network card, and is a communication unit that communicates with a plurality of terminal devices 2 and the like via the network NW.

  The portable storage medium drive 15 is a device that writes information to and reads information from the portable storage medium 150. Examples of the portable storage medium 150 include a USB memory, a CD-R, and a memory card. The customer analysis program may be stored in the portable storage medium 150.

  The customer analysis apparatus 1 further includes an input device 160 for performing an information input operation and a display 170 for displaying an image. The input device 160 is input means such as a keyboard and a mouse, and input information is output to the CPU 10 via the input processing unit 16. The analysis target period designation information may be input via the input device 160.

  The display 170 is image display means such as a liquid crystal display, and displayed image data is output from the CPU 10 to the display via the image processing unit 17. In addition, it can replace with the input device 160 and the display 170, and can also use devices, such as a touchscreen provided with these functions.

  The CPU 10 executes a program stored in the ROM 11 or the HDD 13 or a program read from the portable storage medium 150 by the portable storage medium drive 15. This program includes not only the OS (Operating System) but also the customer analysis program described above. The customer analysis program may be downloaded via the communication processing unit 14.

  When the CPU 10 executes the customer analysis program, a plurality of functions are formed. In the first embodiment, the CPU 10 generates an integrated customer group including similar customer groups according to the analysis target period, compares the generated integrated customer group with the customer group of the department store X to be analyzed, and compares the results. Has a function of outputting. In the second embodiment, the CPU 10 extracts an integrated customer group corresponding to the analysis target period from existing integrated customer groups including a similar customer group, and extracts the extracted integrated customer group and the department store X to be analyzed. It has a function of comparing with a customer group and outputting a comparison result.

(First embodiment)
First, the first embodiment will be described. FIG. 5 is a configuration diagram illustrating a functional configuration example of the customer analysis device 1 according to the first embodiment. FIG. 5 shows an example of functions formed in the CPU 10 and information stored in the HDD 13.

  The CPU 10 performs customer analysis by data clustering. The CPU 10 includes a data input processing unit (acceptance processing unit) 101, a cluster generation unit 102, and a cluster processing unit (output processing unit) 103. The cluster processing unit 103 includes a similarity determination processing unit 103a and a comparison target cluster generation unit 103b.

  The HDD 13 stores cluster generation data 131, an individual company cluster database (DB) 132, and a universal cluster database (DB) 133. The HDD 13 stores a universal cluster candidate database (DB) 134 and an analysis result database (DB) 135.

  The data input processing unit 101 is input with the analysis target period designation information, purchase history data 230, and customer data 231 from the terminal device 2 via the communication processing unit 14. That is, the data input processing unit 101 receives from the terminal device 2 specification (instruction) of the analysis target period, purchase history data 230 and customer data 231 of the analysis target department store X. The data input processing unit 101 outputs the analysis target period designation information to the comparison target cluster generation unit 103b.

  The data input processing unit 101 generates cluster generation data 131 based on the purchase history data 230 and the customer data 231. Here, the “cluster” refers to a bundle of data obtained by classifying customers into groups according to various conditions, that is, customer group information.

  The cluster generation unit 102 generates individual company clusters related to the department stores A to X based on the cluster generation data 131 and stores them in the individual company cluster DB 132. The individual company cluster is a cluster generated for each of the department stores A to X, that is, an individual cluster of the department stores A to X. That is, the cluster generation unit 102 generates a plurality of customer groups based on the purchase history data 230 for each of the department stores A to X.

  The cluster processing unit 103 performs a comparative analysis process of the individual company cluster of the department store X to be analyzed and a universal cluster construction process. The comparison target cluster generation unit 103b generates the individual company cluster (customer group) from the individual company cluster DB 132 based on the purchase history data 230 of the department stores A, B,... To extract. The comparison target cluster generation unit 103b generates a comparison target cluster (integrated customer group) including all the individual company clusters having a predetermined similarity among the extracted individual company clusters. The cluster processing unit 103 executes the comparative analysis process by determining whether there is a similarity between the individual company cluster of the department store X to be analyzed and the comparison target cluster. At this time, the determination process of the similarity relationship (similarity) is performed by the similarity determination processing unit 103a.

  In addition, the cluster processing unit 103 updates the universal cluster DB 133 and the universal cluster candidate DB 134 by performing a universal cluster construction process. The cluster processing unit 103 compares the individual company cluster in the individual company cluster DB 132 with the universal cluster in the universal cluster DB 133, and updates the universal cluster DB 133 according to the comparison result. The universal cluster is an integrated cluster obtained by bundling a plurality of similar company clusters according to a predetermined condition.

  More specifically, the cluster processing unit 103 updates the universal cluster by including an individual company cluster having similarity to the universal cluster in the universal cluster. That is, the cluster processing unit 103 detects a customer group (individual cluster) that is generated based on the purchase history data 230 of the department store X to be analyzed, and has a predetermined similar relationship with the integrated customer group (universal cluster). Update the integrated customer group by including it in the customer group. At this time, the determination process of the similarity relationship (similarity) is performed by the similarity determination processing unit 103a.

  Further, the cluster processing unit 103 compares the individual company cluster in the individual company cluster DB 132 with the universal cluster candidate in the universal cluster candidate DB 134 and updates the universal cluster candidate DB 134 or the universal cluster DB 133 according to the comparison result. The universal cluster candidate is an individual company cluster that has no similarity to any of the existing universal clusters, and is a candidate for a universal cluster that can be a new universal cluster.

  More specifically, the cluster processing unit 103 generates a new universal cluster by integrating individual company clusters having similarity with the universal cluster candidate into the universal cluster candidate. That is, the cluster processing unit 103 generates an integrated customer group (universal cluster) to which a customer group having a predetermined similar relationship belongs among a plurality of customer groups (individual company clusters) generated by the cluster generation unit 102.

  In addition, the cluster processing unit 103 adds an individual company cluster that is not similar to any of the universal cluster candidates to the universal cluster candidate DB 134 as a new universal cluster candidate. Note that the similarity determination processing (similarity) is determined by the similarity determination processing unit 103a.

  FIG. 6 shows an example of the universal cluster DB 133 (FIG. 6A) and the individual company cluster DB 132 (FIG. 6B). The contents of the universal cluster candidate DB 134 are the same as the individual company cluster DB 132.

  The universal clusters “UC-1” and “UC-2” include “number of people” information, “annual purchase amount” information, “age” information, and “original cluster” information. The “number of people” information indicates the number of customers belonging to the universal cluster. For example, “UC-1” and “UC-2” have 100,000 customers.

  The “annual purchase amount” information indicates the ratio of the number of customers when the customers are classified according to the amount of money (“Low” (small), “Middle” (medium), “High” (many)). The “age” information indicates the ratio of the number of customers for each age group (“20s” to “80s”) for each gender (“male” and “female”).

  The “original cluster” information indicates the individual company cluster that is the generation source of the universal cluster. For example, the universal cluster “UC-1” is generated by integrating the individual company clusters “A-1 (2008)” and “B-2 (2010)”, and the universal cluster “UC-2” It is generated by integrating the clusters “C-3 (2010)” and “D-4 (2009)”. Here, the number in parentheses of the individual company cluster indicates the year (“generation year” information described later) of the purchase history data 230 that is the basis when the individual company cluster is generated by the cluster generation unit 102. . That is, “A-1 (2008)” and “B-2 (2010)” are individual company clusters generated based on the purchase history data 230 of the fiscal year 2008 and the fiscal year 2010, respectively.

  The individual company cluster includes “generation year” information, “number of people” information, “annual purchase amount” information, “age” information, and “original cluster” information. The “number of people” information, the “annual purchase amount” information, and the “age” information are as described above.

  The “generation year” information indicates the year of the purchase history data 230 that is based when the individual company cluster is generated by the cluster generation unit 102. For example, the years of the purchase history data 230 from which the individual company clusters “A-1” and “B-2” are generated are 2008 and 2010, respectively.

  The “generation year” information corresponds to the time indicated by the “purchase date” information in the purchase history data 230 that is the basis (that is, the generation source) of the individual company cluster. That is, when generating the individual company cluster, the cluster generation unit 102 generates “generation year” information based on the year of the “purchase date” information (such as “2011”). As will be described later, when the comparison target cluster generation unit 103b generates the comparison target cluster, the comparison target cluster generation unit 103b extracts individual company clusters whose period indicated by the “generation year” information is included in the analysis target period. Instead of “generation year” information, information such as the month and season in which the individual company cluster was generated may be used.

  Thus, the HDD 13 (individual company cluster DB 132) corresponds each customer group (individual company cluster) with the time indicated by the purchase history data 230 that is the basis of each customer group (the year indicated by the “purchase date” information). Add and remember. Therefore, the comparison target cluster generation unit 103b can easily extract individual company clusters corresponding to the specified analysis target period.

  The cluster processing unit 103 compares the generated comparison target cluster with the individual company cluster of the analysis target department store X, and stores the comparison result in the analysis result DB 135. Note that the comparison target cluster is deleted without being stored in the HDD 13 or the like when the comparison analysis process is completed.

  The cluster processing unit 103 reads information indicating the comparison result from the analysis result DB 135 at a predetermined timing, and transmits the information to the terminal device 2. The information indicating the comparison result includes an individual company cluster dissimilar to the comparison target cluster, and a difference between the comparison target cluster similar to the individual company cluster and the individual company cluster. The individual company cluster dissimilar to the comparison target cluster is processed as a characteristic customer group of the department store X to be analyzed.

  That is, the cluster processing unit 103 detects a customer group (individual company cluster) that does not have a predetermined similarity with the integrated customer group (comparison target cluster), and outputs it as a characteristic customer group of the department store X to be analyzed. To do. Further, the cluster processing unit 103 detects and outputs a difference between the integrated customer group and the customer group having a predetermined similarity with the integrated customer group. The output information of the comparison result is used for planning a business strategy at the department store X to be analyzed. Details of the customer analysis program based on the above-described configuration will be described below.

  FIG. 7 is a flowchart of the customer analysis program according to the first embodiment. FIG. 7 shows the flow of one analysis process. In this example, a case where both the process related to the cluster comparative analysis process (steps St4 to St6) and the universal cluster construction process (step St7) are executed is described, but the present invention is not limited to this. Only one may be executed.

  First, the data input processing unit 101 receives the analysis target period designation information (instruction information), purchase history data 230, and customer data 231 from the terminal device 2 (step St1). That is, the data input processing unit 101 accepts specification (instruction) of an analysis target period and purchase history data of a store to be analyzed.

  Next, the data input processing unit 101 generates cluster generation data 131 based on the purchase history data 230 and the customer data 231 (step St2).

  FIG. 8A shows an example of the cluster generation data 131. The cluster generation data 131 includes “customer ID” and “brand” information. “Customer ID” corresponds to the customer ID of the purchase history data 230. The “brand” information corresponds to the “brand name” in the purchase history data 230, and the customer for each brand type (“A” to “C”) within a certain period (eg, within a specific year, within a specific season, etc.). Indicates the number of purchases of.

  For example, for customer ID “0001”, the number of purchases of brands A to C is 3, 2, and 0, respectively, and for customer ID “0002”, the number of purchases of brands A to C is 0 times, 1 time and 2 times.

  Next, the cluster generation unit 102 reads the cluster generation data 131 and generates an individual company cluster based on the cluster generation data 131 (step St3). That is, the cluster generation unit 102 generates an individual company cluster by performing data clustering based on the purchase history data 230 and the customer data 231. At this time, the cluster generation unit 102 generates the above “generation year” information based on the purchase history data 230. The generated individual company cluster is stored in the individual company cluster DB 132.

  Although there is no limitation on the data clustering method, an example is a method of mapping customers (customer IDs) to three-dimensional coordinates based on the cluster generation data 131 shown in FIG.

  FIG. 8B shows an example of the three-dimensional coordinates for cluster generation. In this example, the coordinates for each customer (customer ID) are defined by three coordinate axes A to C corresponding to the number of purchases of brands A to C, respectively.

  For example, for customer ID “0001”, the number of purchases of brands A to C is 3, 2, and 0, respectively, and the coordinates are (3, 2, 0). In addition, since the customer ID “0002” has the number of purchases of the brands A to C being 0 times, 1 time, and 2 times, the coordinates are (0, 1, 2). Similarly, the coordinates of the customer IDs “0003” and “0004” are (2, 2, 1) and (0, 2, 3), respectively.

  The cluster generation unit 102 generates individual company clusters by bundling customer IDs having similar mapped coordinates and classifying the customers into a plurality of groups. For example, customer IDs “0001” and “0003” are classified into the group G1, and customer IDs “0002” and “0004” are classified into the group G2. The individual company cluster DB 132 corresponding to the group G1 indicates a customer group having a strong purchase tendency for the brands A and B, and the individual company cluster DB 132 corresponding to the group G2 indicates a customer group having a strong purchase tendency for the brands B and C.

  As described above, as an example, the cluster generation unit 102 generates individual company clusters according to the purchase trends of the brands A to C for each customer, thereby enabling customer analysis according to the purchase trends. However, the criteria for generating the individual company cluster are not limited to the purchasing tendency of the brands A to C, but may be various information (see FIG. 3) such as age group, residential area, sex, and purchase date. .

  Next, the comparison target cluster generation unit 103b generates a comparison target cluster of the individual company cluster of the analysis target department store X, that is, a comparison target cluster (step St4). The comparison target cluster is generated by integrating all the individual company clusters having similarities among the individual company clusters of other department stores A, B,... Stored in the individual company cluster DB 132.

  FIG. 9 is a flowchart showing the comparison target cluster generation processing. First, the comparison target cluster generation unit 103b extracts an individual company cluster corresponding to the analysis target period from the individual company cluster DB 132 (step St21).

  As described above, the analysis target period designation information indicates a specific period or a predetermined period from the current or past time point. For example, when comparing the information of the customer group of the department store X to be analyzed (individual company cluster) with a long-term trend, a period prior to the past specific period (for example, 2011) is specified as the period to be analyzed. In this case, the comparison target cluster generation unit 103b uses the generation year (“generation year” information) as the basis of the comparison target cluster (the generation source cluster) before the specific time (for example, FY2009, FY2010). , 2011) is extracted.

  Moreover, when comparing the information of the customer group of the department store X to be analyzed with a short-term trend, a period after the past specific period (for example, 2012) is designated as the period to be analyzed. In this case, the comparison target cluster generation unit 103b extracts individual company clusters whose generation year is earlier than the specific time (for example, 2012, 2013) as the basis of the comparison target cluster. The designation of the analysis target period is not limited to the one based on the past time point, and may be one based on the present (for example, “for the past two years”).

  Furthermore, when the information of the customer group of the department store X to be analyzed is compared with a trend at a specific time, a specific period (for example, 2013) is designated as the analysis target period. In this case, the comparison target cluster generation unit 103b extracts an individual company cluster whose generation year is the specific period (for example, 2013) as the basis of the comparison target cluster.

  Next, the comparison target cluster generation unit 103b determines whether there are a plurality of extracted individual company clusters (step St22). When the extracted individual company cluster is 1 or less (No in step St22), the comparison target cluster generation unit 103b ends the process. In this case, since the comparison target cluster is not generated, the processes of steps St5 and St6 in FIG. 7 are not performed.

  When there are a plurality of extracted individual company clusters (Yes in step St22), the comparison target cluster generation unit 103b determines whether there is similarity between the extracted individual company clusters (step St23). If there is no similarity for any combination of individual company clusters (No in step St23), the comparison target cluster generation unit 103b ends the process. In this case, since the comparison target cluster is not generated, the processes of steps St5 and St6 in FIG. 7 are not performed.

  When there is similarity for any combination of individual company clusters (Yes in step St23), the comparison target cluster generation unit 103b merges all individual company clusters that are similar to each other (step St24). Thereby, a comparison target cluster (integrated customer group) including a plurality of individual company clusters (customer groups) similar to each other is generated. In this way, the comparison target cluster generation process is performed. The comparison target cluster is a cluster that is temporarily generated for the comparison analysis process, and is not stored in the HDD 13 or the like. For example, the comparison target cluster is deleted from the cluster processing unit 103 after the comparison analysis process ends.

  In this way, in this embodiment, since the comparison target cluster is generated from the individual company cluster extracted from the individual company cluster DB 132, unlike the second embodiment described later, in the universal cluster DB 133, according to the analysis target period. Customer analysis is possible without a universal cluster. Therefore, for example, the customer analysis apparatus 1 may select and use either the customer analysis method according to the present embodiment or the customer analysis method according to the second embodiment according to the contents of the universal cluster DB 133. In the following, a generation example of a comparison target cluster will be described by taking as an example the case where “2010 fiscal year” is designated as the analysis target period.

  FIG. 10 shows a generation example of the comparison target cluster. FIG. 10A shows a state of extraction of individual company clusters, and FIG. 10B shows a state of generation of comparison target clusters.

  As shown in FIG. 10A, the comparison target cluster generation unit 103b determines that the individual company cluster in the individual company cluster DB 132 is based on the “generation year” information and the individual company cluster in 2010 and other than the 2010 fiscal year. Into individual company clusters. In this example, “B-2 (2010)” of department store B, “C-3 (2010)” of department store C, and “D-4 (2010)” of department store D are shown as individual company clusters in 2010. Has been.

  In addition, “A-1 (2008)” of department store A and “E-5 (2009)” of department store E are shown as individual company clusters other than 2010. The comparison target cluster generation unit 103b extracts “B-2 (2010)”, “C-3 (2010)”, and “D-4 (2010)” as individual company clusters corresponding to the analysis target period. In the following description, the numbers in parentheses assigned to the cluster numbers (A-1, E-5, etc.) indicate “generation year” information.

  Next, as illustrated in FIG. 10B, the comparison target cluster generation unit 103 b determines the presence or absence of similarity for all combinations of the extracted individual company clusters, and “B-2 (2010)” that are similar to each other. ”And“ C-3 (2010) ”are merged (integrated). Thereby, the comparison target cluster “CC-1” is generated. Since “D-4 (2010)” is not similar to either “B-2 (2010)” or “C-3 (2010)”, it is not used to generate the comparison target cluster. Moreover, although the determination method of the presence or absence of the similarity between clusters is not limited, For example, as described below, similarity may be determined by comparing one or more indices of each cluster.

  FIG. 11 shows an example of an index used for the similarity determination process. Examples of the index include an RFM value, customer attributes, and purchase details.

  The RFM value includes “Recency”, “Frequency”, and “Monetary”. “Recency” indicates the number of days that have elapsed since the last purchase by the customer. “Frequency” indicates the purchase frequency of the customer. “Monetary” indicates the purchase amount of the customer.

  The median or average value of the distribution of the number of customers for each of the three index values of “Recency”, “Frequency”, and “Monetary” (hereinafter referred to as “index value”) is compared using three-dimensional coordinates. Thus, the similarity of clusters can be determined.

  The customer attributes include “age group”, “gender ratio”, and “residential area distribution”. The “age group”, “sex”, and “resident area” are acquired from the “age” information, the “sex” information, and the “zip code” in the customer data 231, respectively.

  As the index value of “age group”, for example, customer age distribution, median value, and average value are used. Examples of the “sex” index value include a male-female ratio and the number of people. Examples of the index value of “resident area” include the number of people and the distribution for each area.

  The purchase contents of the customer include “brand”, “amount”, “visit frequency”, and “number of customers”. “Brand”, “Price”, “Visit frequency”, and “Number of customers” are acquired from the purchase history data 230.

  Examples of the “brand” index value include a characteristic brand purchased by a customer and a ranking of the brand. Examples of the “money amount” index value include the annual purchase amount of the customer, the purchase amount for each visit to the store, and the purchase amount for each day of the week. Examples of the index value of “visit frequency” include the number of visits by customers in one week and the number of visits in one month. As an index value of “number of customers”, for example, the number of customers who visit each customer day of the week can be cited.

  Thus, various index values can be used to determine the similarity between clusters. In the following example, the case where the change characteristics of two index values are used for determination will be described as an example.

  FIG. 12 is a graph showing a change characteristic of the index. FIG. 12 shows the change characteristics of the index values x and y of the clusters “A-1”, “B-1”, and “C-1” when the period shifts from t1 to t2. The periods t1 and t2 may be a specific month (for example, January, June, etc.), or before and after the same event (promotion of sales, implementation of sales floor measures, etc.) of department stores A to X. It may be a period.

  “A (t1) -1”, “B (t1) -1”, and “C (t1) -1” are indices of “A-1”, “B-1”, and “C-1” in the period t1. Indicates the coordinates (x, y) of the value. “A (t2) -1”, “B (t2) -1”, and “C (t2) -1” are “A-1”, “B-1”, and “C-1” in the period t2. Indicates the coordinates (x, y) of the index value. For example, the index value (x, y) of “A-1” changes from (xa1, ya1) to (xa2, ya2), and the index value (x, y) of “B-1” is (xb1, It changes from yb1) to (xb2, yb2).

  The comparison target cluster generation unit 103b compares the change characteristics of the index values x and y of each cluster, and determines that the clusters showing the same change tendency as the comparison result have similarity. In this example, since the index values x and y of “A-1” and “B-1” show the same change tendency, “A-1” and “B-1” are similar clusters. Determined. Hereinafter, the similarity determination means (1) to (3) will be described with examples.

(Determination means (1))
FIG. 13A shows an example of determination of cluster similarity by the determination unit (1). The determination means (1) detects whether the change direction of each index value x, y is “increase”, “decrease”, or “no change” when the period changes from t1 to t2. When the change directions of both index values x and y coincide, it is determined that the clusters have similarity. In FIGS. 13A to 13C, “No.” indicates a number for distinguishing determination examples.

  In the case of the example in FIG. 12, the index value xa2 of “A (t2) −1” is larger than the index value xa1 of “A (t1) −1”, and therefore the index value x of “A-1” is “increased”. Indicates. Further, since the index value ya2 of “A (t2) −1” is larger than the index value ya1 of “A (t1) −1”, the index value y of “A-1” indicates “increase”. On the other hand, the index values x and y of “C-1” both indicate “decrease”. When the rate of change of the index values x and y is less than 5 (%), for example, the index values x and y are determined as “no change”.

  “No. 1” indicates a determination example of the clusters “A-1” and “B-1”. Each index value x, y of “A-1” indicates “increase”, and each index value x, y of “B-1” also indicates “increase”. That is, the change directions of “A-1” and “B-1” are the same. Therefore, the similarity between “A-1” and “B-1” is determined as “present”.

  “No. 2” indicates a determination example of the clusters “A-1” and “C-1”. Each index value x, y of “A-1” indicates “increase”, and each index value x, y of “C-1” indicates “decrease”. That is, the change directions of “A-1” and “C-1” do not match. Therefore, the similarity between “A-1” and “C-1” is determined as “none”.

  “No. 3” indicates a determination example of the clusters “D-1” and “E-1”. Each index value x, y of “D-1” indicates “increase” and “decrease”, respectively, and each index value x, y of “E-1” indicates “decrease” and “increase”, respectively. That is, the change directions of “D-1” and “E-1” do not match. Therefore, the similarity between “D-1” and “E-1” is determined as “none”.

  “No. 4” indicates a determination example of the clusters “UC-2” and “X-1”. Each index value x, y of “UC-2” indicates “increase” and “no change”, and each index value x, y of “X-1” indicates “increase” and “no change”, respectively. Show. That is, “UC-2” and “X-1” change in the same direction. Therefore, the similarity between “UC-2” and “X-1” is determined as “present”.

  In this way, the determination means (1) facilitates similarity only by detecting whether the change direction of the index values x and y is “increase”, “decrease”, or “no change”. Can be determined.

(Determination means (2))
FIG. 13B shows a determination example of cluster similarity by the determination unit (2). In the determination means (2), the change directions (increase or decrease) of the index values x and y coincide, and the ratios Δx and Δy of the difference between the index values x and y between the clusters in the periods t1 and t2 are respectively predetermined. If the threshold value THx is less than THy, it is determined that the clusters have similarity. Here, the ratios Δx and Δy are calculated based on one of the clusters to be compared.

  In the example of FIG. 12, when “A-1” and “B-1” are compared, the ratio Δx (t1) of the difference between the index values x in the period t1 is | xb1-xa1 | × 100 / xa1 × 100. Yes, the difference ratio Δx (t2) of the index value x in the period t2 is | xb2-xa2 | / xa2 × 100. Further, the difference ratio Δy (t1) of the index value y in the period t1 is | yb1-ya1 | / ya1 × 100, and the difference ratio Δy (t2) of the index value y in the period t2 is | yb2-ya2. | / Ya2 × 100. Note that the change characteristics (increase) of the index values x and y coincide. Note that the ratios Δx (t1), Δx (t2), Δy (t1), and Δy (t2) are calculated based on the cluster “A-1”.

Δx (t1) <THx (1)
Δx (t2) <THx (2)
Δy (t1) <THy (3)
Δy (t2) <THy (4)

  Therefore, if all of the above formulas (1) to (4) hold, it is determined that “A-1” and “B-1” have similarity. In the following determination example, THx = 20 (%) and THy = 5 (%).

  “No. 5” indicates a determination example of the clusters “A-1” and “B-1”. Each index value x, y of “A-1” indicates “increase”, and each index value x, y of “B-1” also indicates “increase”. That is, the change directions of “A-1” and “B-1” are the same.

  The ratios Δx (t1), Δx (t2), Δy (t1), and Δy (t2) of the difference between the index values x and y of “A-1” and “B-1” are expressed by the above equation (1). Each of (4) is satisfied. Therefore, the similarity between “A-1” and “B-1” is determined as “present”.

  “No. 6” indicates a determination example of the clusters “C-1” and “D-1”. Each index value x, y of “C-1” indicates “increase” and “decrease”, respectively, and each index value x, y of “D-1” also indicates “increase” and “decrease”, respectively. That is, the change directions of “C-1” and “D-1” are the same.

  Further, the difference ratios Δx (t1) and Δy (t1) between the index values x and y of “C-1” and “D-1” satisfy the above formulas (1) and (3), respectively. Ratios Δx (t2) and Δy (t2) do not satisfy the expressions (2) and (4), respectively. Therefore, the similarity between “C-1” and “D-1” is determined as “none”.

  Thus, the determination means (2) is more accurate than the determination means (1) based not only on the direction of change of the index values x and y but also on the ratios Δx and Δy of the index values x and y between the clusters. Similarity can be determined.

(Determination means (3))
FIG. 13C shows an example of determination of cluster similarity by the determination unit (3). In the determination means (3), when the time periods change from t1 to t2, the change rates Rx and Ry of the index values x and y between the clusters ΔRx and ΔRy are less than the predetermined thresholds THrx and THry, respectively. The clusters are determined to have similarity. In the determination means (3), the coincidence of the change directions of the index values x and y is a similarity condition.

  In the example of FIG. 12, when “A-1” and “B-1” are compared, the rate of change Rx (A) of the index value x of “A-1” is (xa2−xa1) × 100 / xa1. Yes, the rate of change Rx (B) of the index value x of “B-1” is (xb2−xb1) × 100 / xb1. Further, the rate of change Ry (A) of the index value y of “A-1” is (ya2−ya1) × 100 / ya1, and the rate of change Ry (B) of the index value y of “B-1” is (Yb2-yb1) × 100 / yb1.

ΔRx = Rx (A) −Rx (B) <THrx (5)
ΔRy = Ry (A) −Ry (B) <THry (6)

  Therefore, if all of the above formulas (5) and (6) hold, “A-1” and “B-1” are determined to have similarity. In the following determination example, THrx = 20 (%) and THry = 5 (%).

  “No. 7” indicates a determination example of the clusters “A-1” and “B-1”. Each index value x, y of “A-1” indicates “increase”, and each index value x, y of “B-1” also indicates “increase”. That is, the change directions of “A-1” and “B-1” are the same.

  Further, since the change rates Rx of the index values x of “A-1” and “B-1” are 10 (%) and 20 (%), respectively, the difference ΔRx is 10 (%). On the other hand, since the change rates Ry of the index values y of “A-1” and “B-1” are 10 (%) and 13 (%), respectively, the difference ΔRy is 3 (%). Therefore, the differences ΔRx and ΔRy satisfy the above equations (5) and (6), respectively. Therefore, the similarity between “A-1” and “B-1” is determined as “present”.

  “No. 8” indicates a determination example of the clusters “C-1” and “D-1”. Each index value x, y of “C-1” indicates “increase”, and each index value x, y of “D-1” also indicates “increase”. That is, the change directions of “C-1” and “D-1” are the same.

  Further, since the change rates Rx of the index values x of “C-1” and “D-1” are both 10 (%), the difference ΔRx is 0 (%). On the other hand, since the change rates Ry of the index values y of “C-1” and “D-1” are 35 (%) and 20 (%), respectively, the difference ΔRy is 15 (%). Therefore, the difference ΔRx satisfies the above equation (5), but the difference ΔRy does not satisfy the above equation (6). Therefore, the similarity between “C-1” and “D-1” is determined as “none”.

  As described above, the determination unit (3) determines not only the change direction of the index values x and y but also the determination unit (1) based on the differences ΔRx and ΔRy between the change rates Rx and Ry of the index values x and y between clusters. ) Similarity can be determined with higher accuracy.

  In addition, when the comparison target cluster generation unit 103 b merges similar individual company clusters to generate the comparison target cluster “CC-1”, the comparison target cluster generation unit 103 b updates the numerical values of the information included in the individual company cluster by recalculation. . Hereinafter, a calculation example will be described by taking as an example the case of merging the individual company clusters “B-2 (2010)” and “C-3 (2010)” shown in FIG.

  FIG. 14 shows an example of the comparison target cluster. The comparison target cluster “CC-1” is generated by merging the individual company clusters “B-2 (2010)” and “C-3 (2010)”. Hereinafter, calculation examples of the “number of people” information and the “annual purchase amount” information will be described.

  The “number of people” of “B-3 (2010)” is 5000 people, and the “number of people” of “C-3 (2010)” is 10,000 people. Therefore, the “number of people” of “CC-1” is calculated as 15000 people from the sum of 5000 people and 10,000 people.

  In addition, since “Low” in the “annual purchase amount” information of “B-3 (2010)” is 5 (%), the number of customers is 250 (= 5000 (refer to “number”) × 5 (%)). Since “Low” in the “annual purchase amount” information of “C-3 (2010)” is 5 (%), the number of customers is 500 (= 10000 (refer to “number”) × 5 ( %)).

  Therefore, “Low” of the “annual purchase amount” information of “CC-1” is the number of people (250 people) of “B-3 (2010)” and the number of people (500 people of “C-3 (2010)”). ) Is divided by the “number of people” (15000 people) of “CC-1” to be calculated as 5 (%). As described above, each piece of information of the comparison target cluster is updated by recalculating the numerical values of the information of a plurality of individual company clusters that are similar to each other.

  Referring to FIG. 7 again, the cluster processing unit 103 performs a comparative analysis process between the comparison target cluster generated by the comparison target cluster generation unit 103b and the individual company cluster of the analysis target department store X (step St5). Thereby, the analysis result (comparison result) is stored in the analysis result DB 135.

  FIG. 15A is a flowchart of cluster comparison analysis processing in the first embodiment. In FIG. 15A, an example of the process of step St51 is indicated by reference numeral C51.

  In this example, the individual company cluster DB 132 includes the existing individual company clusters “A-1 (2008)”, “B-2 (2010)”, “C-3 (2010)”, and the department store to be analyzed. X individual company clusters “X-1” to “X-3” are stored. Here, the comparison target cluster generation unit 103b uses the comparison target cluster “A-1 (2008)”, “B-2 (2010)”, “C-3 (2010)”, and the like. Assume that CC-1 "to" CC-3 "are generated.

  First, the similarity determination processing unit 103a compares the individual company clusters “X-1” to “X-3” with the comparison target clusters “CC-1” to “CC-3” (Step St51). At this time, the similarity determination processing unit 103a, as indicated by reference numeral C51, all of the individual company clusters “X-1” to “X-3” and the comparison target clusters “CC-1” to “CC-3”. The combination is determined whether there is similarity.

  Next, as a result of the comparison, if the individual company cluster is not similar to any of the comparison target clusters (No in step St52), the cluster processing unit 103 stores the individual company cluster as a characteristic customer group in the analysis result DB 135. (Step St53). That is, the cluster processing unit 103 generates a customer group (individual company cluster) that is generated based on the purchase history data 230 of the department store X to be analyzed and does not have a predetermined similar relationship with the integrated customer group (comparison target cluster). It detects and outputs as a characteristic customer group of the department store X to be analyzed.

  If the individual company cluster is similar to any of the comparison target clusters (Yes in step St52), the cluster processing unit 103 detects a difference between the individual company cluster and the comparison target cluster (step St58). Information indicating the difference between the individual company cluster and the comparison target cluster is stored in the analysis result DB 135. In this way, the cluster comparison analysis process is performed.

  As described above, the cluster processing unit 103 includes an integrated customer group (comparison target cluster) including all customer groups having a predetermined similarity relationship, and a customer group generated based on the purchase history data 230 of the department store X to be analyzed. Outputs the comparison result with (individual company cluster). As described above, the comparison result (customer analysis result) is temporarily stored in the analysis result DB 135.

  Referring to FIG. 7 again, the cluster processing unit 103 reads out the analysis results (comparison results) obtained in the above steps St53 and St58 from the analysis result DB 135 and outputs them to the terminal device 2 via the communication processing unit 14 ( Send) (step St6). In the department stores A to X, the analysis result is taken out from the terminal device 2 and used for planning a business strategy. Next, the cluster processing unit 103 executes a universal cluster construction process (step St7).

  FIG. 15B is a flowchart of a universal cluster construction process. In this process, at least one of the universal cluster DB 133 and the universal cluster candidate DB 134 is updated.

  Moreover, FIG. 15B shows the example of a process of step St11, St14, St17, St19 by code | symbol C1-C4. In this example, universal clusters “UC-1” to “UC-3” are stored in the universal cluster DB 133, and universal cluster candidates “A-2”, “B-3”, and “D-4” is stored. The individual company cluster DB 132 stores individual company clusters “X-1” to “X-3” of the department store X to be analyzed.

  First, the similarity determination processing unit 103a performs similarity determination by comparing the individual company clusters “X-1” to “X-3” with the universal clusters “UC-1” to “UC-3” (step St11). . At this time, the similarity determination processing unit 103a performs all combinations of the individual company clusters “X-1” to “X-3” and the universal clusters “UC-1” to “UC-3” as indicated by reference numeral C1. Whether or not there is similarity is determined. The similarity determination means is as described with reference to FIG.

  As a result of the similarity determination, when the individual company cluster is not similar to any of the universal clusters (No in step St12), the cluster processing unit 103 compares the individual company cluster with the universal cluster candidate by the similarity determination processing unit 103a ( Step St14). Here, when the individual company cluster “X-3” is not similar to “UC-1” to “UC-3”, “X-3” is stored in the universal cluster candidate DB 134 as indicated by reference symbol C2. Similarity is determined among all the universal cluster candidates “A-2”, “B-3”, and “D-4”. The similarity determination means is as described with reference to FIG.

  As a result of the comparison, if the individual company cluster is not similar to any of the universal cluster candidates (No in step St15), the cluster processing unit 103 adds the individual company cluster as a universal cluster candidate to the universal cluster candidate DB 134 (step St16), the process is terminated. That is, the cluster processing unit 103 updates the universal cluster candidate DB 134 by detecting a customer group that does not have a similar relationship with any of the universal cluster candidates and including it in the universal cluster candidate. As a result, a customer group cluster that can become a new integrated customer group (universal cluster) is added to the universal cluster candidate DB 134 as a new universal cluster candidate.

  In addition, if the comparison result shows that the individual company cluster is similar to any of the universal cluster candidates (Yes in step St15), the cluster processing unit 103 merges the individual company cluster with the universal cluster candidate to create a new one. A universal cluster is generated (step St17). For example, when the individual company cluster “X-3” and the universal cluster candidate “B-3” are similar, “X-3” and “B-3” are integrated as shown by reference numeral C4, and a new universal A cluster “UC-4” is generated. The cluster processing unit 103 adds the generated universal cluster to the universal cluster DB 133 and ends the process.

  As described above, the cluster processing unit 103 detects a customer group (individual company cluster) having a predetermined similarity relationship with a candidate for an integrated customer group (universal cluster). The cluster processing unit 103 generates a new integrated customer group by including the detected customer group in an integrated customer group candidate (universal cluster candidate). Therefore, as customer analysis is repeated, the contents of the universal cluster DB 133 are enriched, and highly accurate customer analysis becomes possible.

  Note that “UC-1” to “UC-3” already stored in the universal cluster DB 133 are also generated through the processing of step St17. That is, “UC-1” to “UC-3” to be compared with the individual company clusters “X-1” to “X-3” of the department store X to be analyzed in Step St11 were executed in the past customer analysis. This is generated by the processing of step St17.

  If the individual company cluster is similar to any of the universal clusters (Yes in step St12), the cluster processing unit 103 updates the universal cluster DB 133 by merging the individual company cluster and the universal cluster (step St19), the process is terminated. For example, when the individual company cluster “X-1” and the universal cluster “UC-1” are similar, as indicated by reference numeral C3, “X-1” and “UC-1” are integrated, and “UC-1” '"Is generated. If the individual company cluster “X-2” and the universal cluster “UC-2” are similar, “X-2” and “UC-2” are integrated to generate “UC-2 ′”. The generated “UC-1 ′” and “UC-2 ′” are stored in the universal cluster DB 133.

  As described above, the cluster processing unit 103 determines that the customer group (individual cluster) of the analysis target department store X generated based on the purchase history data 230 of the analysis target department store X is an integrated customer group (universal cluster). It detects that it has a predetermined similarity relationship. The cluster processing unit 103 updates the integrated customer group by including the customer group in the integrated customer group. The cluster processing unit 103 uses the updated integrated customer group for further department store customer analysis. Therefore, as customer analysis is performed, the contents of the universal cluster DB 133 are enriched, and highly accurate customer analysis is possible.

  In this way, the universal cluster construction process is performed. After the universal cluster construction process, the customer analysis process ends.

  FIG. 16 shows an output example of the analysis result obtained by the comparative analysis process. In this example, the individual company cluster DB 132 stores individual company clusters “X-1” to “X-3” of the department store X to be analyzed. “X-1” and “X-2” are similar to the comparison target clusters “CC-1” and “CC-2”, respectively, and “X-3” is “CC-1” and “CC-2”. -2 "has no similarity.

  Therefore, “X-3” is stored in the analysis result DB 135 as a characteristic customer group of the department store X to be analyzed. Information indicating the difference between “X-1” and “CC-1” and information indicating the difference between “X-2” and “CC-2” are also stored in the analysis result DB 135.

  FIG. 17 shows an output example of the difference between the individual company cluster and the comparison target cluster. In this example, information on the difference between the universal cluster “CC-1” and the individual company cluster “X-1” is shown for each “item”.

  Examples of the “item” of the difference include, for example, “purchase details” of the customer (product category, etc.), “amount / frequency of visit” (number of each year, month, day of the week, etc.), “age group” (age distribution) Etc.), “gender” (gender ratio, etc.), and “residential area” (residential area distribution, etc.). In this way, the difference between the individual company cluster and the universal cluster similar to the individual company cluster is output as an analysis result, and the customer group of the department store X to be analyzed is compared with other similar customer groups. Detailed information can be obtained. Below, the customer analysis example 1 is given and demonstrated about utilization of an analysis result.

(Customer analysis example 1)
FIG. 18 shows a cluster before the customer analysis processing in the customer analysis example 1.

  The universal cluster DB 133 stores universal clusters “UC-1” to “UC-10”, and the universal cluster candidate DB 134 stores universal cluster candidates “A-4” (generated from the individual company cluster of department store A) and “ "C-5" (generated from a cluster of individual companies of department store C) is stored. “UC-1” to “UC-10” include individual company clusters included in the universal cluster (see “original cluster” information in FIG. 6A) and “generation year” information (parentheses) of the individual company clusters. (See the numbers in the table). For example, “UC-1” includes private company clusters “A-1 (2010)” and “B-1 (2011)”, and “UC-2” is private company cluster “B-2 (2011)”. And “C-6 (2012)”.

  The individual company cluster DB 132 includes individual company clusters “A-1” to “A-7” of the department store A, individual company clusters “B-1” to “B-5” of the department store B, and an individual of the department store C. The company clusters “C-1” to “C-10” are stored. The “generation year” information for each individual company cluster is shown in parentheses.

  Further, purchase history data 230 and customer data 231 are acquired from the terminal device 2 of the department store X to be analyzed. The data amounts of the purchase history data 230 and the customer data 231 are, for example, one year in 2012.

  FIG. 19A shows individual company clusters “X-1 (2012)” to “X-5 (2012)” of the department store X to be analyzed. “X-1 (2012)” to “X-5 (2012)” are generated from the purchase history data 230 and the customer data 231 and stored in the individual company cluster DB 132. The “generation year” information of “X-1 (2012)” to “X-5 (2012)” indicates “2012”.

  “X-1 (2012)” to “X-5 (2012)” are generated based on, for example, purchase contents of 10,000 customers. For this reason, the total number of customers to be analyzed is 50,000.

  FIG. 19B shows the comparison target clusters “CC-1” to “CC-6” generated from the individual company clusters of the department stores A to C when the cluster comparative analysis processing is performed. In this example, it is assumed that two years of 2010 and 2012 are designated as the analysis target period. Therefore, the “generation year” information of the individual company clusters of the department stores A to C that are the generation sources of “CC-1” to “CC-6” indicates at least one of “2010” and “2012” (dotted line frame See inside).

  In “CC-1” to “CC-6”, the individual company cluster included in the comparison target cluster and the “generation year” information (see numbers in parentheses) of the individual company cluster are appended respectively. For example, “CC-1” includes private company clusters “A-2 (2010)” and “C-8 (2012)”, and “CC-2” is private company cluster “A-3 (2010)”. And “C-7 (2012)”.

  Among the comparison target clusters, “CC-1” to “CC-5” have the same individual cluster as the generation source cluster (see “existing cluster”). Here, the existing universal cluster is shown below the symbol “=”. For example, “CC-1” and “CC-2” have the same individual cluster of the generation source as “UC-9” and “UC-7”, respectively.

  “CC-6” is a new cluster (see “new cluster”) in which the individual company cluster of the generation source is not in common with the existing universal cluster. “CC-6” includes individual company clusters “A-1 (2010)” and “C-5 (2012)”.

  “C-5 (2012)” is determined to be similar to “A-1 (2010)”, while “UC-1” including “A-1 (2010)”. It is determined that there is no similarity between. This is because “UC-1” includes not only “A-1 (2010)” but also “B-1 (2011)”, so it does not have the same characteristics as “A-1 (2010)”. Because there is no.

  As described above, in this embodiment, the comparison target cluster generated from the individual company cluster corresponding to the analysis target period is used as the comparison target of the individual company cluster of the department store X to be analyzed, which is different from the existing universal cluster. Comparison analysis with new clusters is possible.

  FIG. 20 shows individual company clusters and comparison target clusters having similarities. In FIG. 20, individual company clusters and comparison target clusters that are similar to each other are shown in association with each other by arrows. For example, “X-1 (2012)” and “X-2 (2012)” are similar to “CC-1” and “CC-6”, respectively.

  The individual company clusters and the comparison target clusters having similarities are compared with each other as described with reference to step St58 of FIG. 15A, and the difference is output as a comparison result (analysis result). Here, the information indicating the difference is as described with reference to FIG.

  In the department store X, information indicating the difference regarding “X-1 (2012)” to “X-5 (2012)” is used for, for example, evaluation of the strength of the customer base of the department store X (so-called benchmark evaluation). Furthermore, department store X analyzes the results of the evaluation and considers measures to increase sales by securing exclusive customer groups from “X-1 (2012)” to “X-5 (2012)”. Is done. As such measures, for example, the customer group finds a product of a brand purchased at another department store A, B, C,..., But not purchased at department store X, and reviews the product or sales floor. Things to do (product measures / sales space measures).

  In the universal cluster construction process, “UC-1” to “UC-10” are the individual company clusters “X-1 (2012)” to “X” as described with reference to step St19 in FIG. 15B. −5 (2012) ”, the similarity is determined, and is updated by being merged with the individual company cluster having the similarity.

As described above, the customer analysis program according to the first embodiment causes the computer (CPU) 10 to execute the following processing.
Process (1): An instruction to designate the period as an analysis target period, or an instruction to set a predetermined period from the current or past time point as an analysis target period, and purchase history data 230 of the department store (store) X to be analyzed; Accept.
Process (2): A predetermined similarity relationship among customer groups (individual company clusters) generated based on purchase history data 230 of a plurality of department stores (stores) A, B,... A comparison result between an integrated customer group (comparison target cluster) including all customer groups having a customer group and a customer group (individual company cluster) generated based on the purchase history data 230 of the department store (store) X to be analyzed To do.

  According to the customer analysis program according to the embodiment, the comparison result obtained by comparing the customer group of the analysis target store X with the integrated customer group to which a similar customer group in the specified analysis target period belongs is output. It is possible to analyze customers taking into account the trend of the season.

In the customer analysis method according to the first embodiment, the computer (CPU) 10 executes the following steps.
Step (1): An instruction to designate a period as an analysis target period, or an instruction to set a predetermined period from the current or past time point as an analysis target period, and purchase history data 230 of an analysis target department store (store) X; Accept.
Step (2): a predetermined similarity relationship among customer groups (individual company clusters) generated based on purchase history data 230 of a plurality of department stores (stores) A, B,... A comparison result between an integrated customer group (comparison target cluster) including all customer groups having a customer group and a customer group (individual company cluster) generated based on the purchase history data 230 of the department store (store) X to be analyzed To do.

  Since the customer analysis method according to the first embodiment includes the same configuration as that of the customer analysis program according to the first embodiment, the same effects as those described above can be obtained.

  Further, the customer analysis device 1 according to the first embodiment includes a reception processing unit (data input processing unit) 101 and an output processing unit (cluster processing unit) 103. The reception processing unit 101 includes an instruction to specify the period as an analysis target period, or an instruction to set a predetermined period from the current or past time point as an analysis target period, and purchase history data of the analysis target store (department store) X Accept.

  The output processing unit 103 has a predetermined similarity among customer groups (individual company clusters) generated based on purchase history data 230 of a plurality of stores (department stores) A, B,... A comparison result between the integrated customer group (universal cluster) including all the customer groups having the relationship and the customer group (individual company cluster) generated based on the purchase history data of the store X to be analyzed is output.

  Since the customer analysis device 1 according to the first embodiment includes the same configuration as the customer analysis program according to the first embodiment, the same effects as those described above can be obtained.

(Second embodiment)
Next, a second embodiment will be described. In the second embodiment, the individual company cluster of the department store X to be analyzed is compared with the universal cluster extracted from the universal cluster DB 133 according to the analysis target period, and the comparison result is output.

  FIG. 21 is a configuration diagram illustrating a functional configuration example of the customer analysis device 1 according to the second embodiment. FIG. 21 shows an example of functions formed in the CPU 10 and information stored in the HDD 13. In FIG. 21, the same components as those in FIG. 5 are denoted by the same reference numerals, and the description thereof is omitted.

  The CPU 10 includes a data input processing unit (acceptance processing unit) 101, a cluster generation unit 102, and a cluster processing unit (output processing unit) 103. The cluster processing unit 103 includes a similarity determination processing unit 103a and a universal cluster extraction unit 103c, and performs cluster comparison analysis processing and universal cluster construction processing. The HDD (storage unit) 13 stores cluster generation data 131, an individual company cluster DB 132, a universal cluster DB 133, a universal cluster candidate DB 134, and an analysis result DB 135.

  The data input processing unit 101 is input with the analysis target period designation information, purchase history data 230, and customer data 231 from the terminal device 2 via the communication processing unit 14. That is, the data input processing unit 101 receives from the terminal device 2 specification (instruction) of the analysis target period, purchase history data 230 and customer data 231 of the analysis target department store X. The data input processing unit 101 outputs the analysis target period designation information to the universal cluster extraction unit 103c.

  The universal cluster extraction unit 103c extracts a universal cluster to be compared with the individual company cluster of the department store X to be analyzed from the universal cluster DB 133 based on the analysis period specified (instructed). At this time, the universal cluster extraction unit 103c refers to the “original cluster” information and the “generation year” information (see FIG. 6) indicating the individual company clusters included in the universal cluster, so that the individual companies corresponding to the analysis target period. Extract universal clusters that contain only clusters.

  As described above, the HDD 13 indicates each customer group (individual company cluster) included in the integrated customer group (universal cluster) as indicated by the purchase history data 230 based on each customer group ("purchase date" information). Are stored in association with each other. For this reason, the universal cluster extraction unit 103c can easily extract the universal cluster from the universal cluster DB 133 by referring to the “original cluster” information and the “generation year” information. That is, the universal cluster extraction unit 103c refers to the HDD 13 and extracts an integrated customer group based only on the purchase history data 230 of the period included in the specified analysis target period from the integrated customer group (universal cluster). .

  In the present embodiment, the cluster processing unit 103 performs a cluster comparison analysis process by comparing the extracted universal cluster and the individual company cluster of the department store X to be analyzed. The cluster processing unit 103 stores the comparison result in the analysis result DB 135. The cluster processing unit 103 stores information indicating the comparison result in the analysis result DB 135, reads it from the analysis result DB 135 at a predetermined timing, and transmits it to the terminal device 2.

  Information indicating the comparison result includes all extracted universal clusters that are dissimilar to the individual company cluster, and the difference between the universal cluster similar to the individual company cluster and the individual company cluster. Individual company clusters that are dissimilar to all the extracted universal clusters are processed as a characteristic customer group of the department store X to be analyzed. The information of the comparison result is used for planning a business strategy at the department store X to be analyzed.

  As will be described later, each time the customer analysis is performed, the cluster processing unit 103 updates the universal cluster DB 133 by performing the universal cluster construction process described above. For this reason, the customer analysis device 1 can perform highly accurate customer analysis as the customer analysis is repeated. Details of the customer analysis program based on the above configuration will be described below.

  FIG. 22 is a flowchart of the customer analysis program according to the second embodiment. FIG. 22 shows the flow of one analysis process. In this example, a case where both the process related to the cluster comparative analysis process (steps St34 to St36) and the universal cluster construction process (step St37) are executed is not limited to this. Only one may be executed.

  First, the data input processing unit 101 receives the analysis target period designation information, purchase history data 230, and customer data 231 from the terminal device 2 (step St31). That is, the data input processing unit 101 receives the analysis target period designation (instruction) and the purchase history data 230 of the analysis target department store X.

  Next, the data input processing unit 101 generates cluster generation data 131 based on the purchase history data 230 and the customer data 231 (step St32). Next, the cluster generation unit 102 reads the cluster generation data 131 and generates an individual company cluster based on the cluster generation data 131 (step St33). The generation process of the cluster generation data 131 and the generation process of the individual company cluster are as described with reference to FIG.

  Next, the universal cluster extraction unit 103c extracts a universal cluster from the universal cluster DB 133 based on the analysis target period designation information input from the data input processing unit 101 (step St34). Hereinafter, the extraction process of the universal cluster will be described with an extraction example.

  FIG. 23 shows an example of universal cluster extraction. In this example, it is assumed that universal clusters “UC-1” to “U-4” are stored in the universal cluster DB 133.

  “UC-1” includes individual company clusters “A-1 (2008)” and “B-2 (2010)”, and “UC-2” includes individual company clusters “C-3 (2011)” and “ D-4 (2009) ". “UC-3” includes individual company clusters “B-3 (2010)” and “E-2 (2013)”, and “UC-4” includes individual company clusters “E-3 (2013)” and “ F-1 (2013) ".

  “Extraction example (1)” indicates a case where a period “before 2010” is designated as the analysis target period. In this case, the universal cluster extraction unit 103c extracts a universal cluster indicating a year (2010, 2009, 2008,...) Whose “generation year” information is “before 2010”. Therefore, in the extraction example (1), “UC-1” based on the individual company clusters in the fiscal years 2008 and 2010 is extracted.

  “Extraction example (2)” indicates a case where a period “2010 and after” is designated as the analysis target period. In this case, the universal cluster extraction unit 103c extracts a universal cluster indicating a year (2010, 2011, 2012,...) Whose “generation year” information is “2010 or later”. Therefore, in the extraction example (2), “UC-3” based on the individual company clusters of 2010 and 2013 and “UC-4” based on the individual company clusters of 2013 are extracted.

  In the extraction example (3), it is assumed that the period “2009-2011” is designated as the analysis target period. In this case, the universal cluster extraction unit 103c extracts universal clusters whose “generation year” information indicates the years “2009” to “2011”. Therefore, in the extraction example (3), “UC-2” based on the individual company clusters in the 2011 and 2009 years is extracted.

  As described above, the universal cluster extraction unit 103c extracts, from the universal cluster DB 133, a universal cluster in which the period indicated by the “generation year” information includes only individual company clusters included in the specified analysis target period. Therefore, the customer analysis device 1 can perform customer analysis in consideration of the trend of the time.

  Next, the cluster processing unit 103 performs a comparative analysis process between the universal cluster extracted by the universal cluster extraction unit 103c and the individual company cluster of the department store X to be analyzed (step St35). Thereby, the analysis result (comparison result) is stored in the analysis result DB 135.

  FIG. 24 is a flowchart of the cluster comparison analysis process in the second embodiment. In FIG. 24, an example of the process in step St41 is indicated by reference numeral C11.

  In this example, the individual company cluster DB 132 stores individual company clusters “X-1” to “X-3” of the department store X to be analyzed. In addition, universal clusters “UC-1” to “UC-10” are stored in the universal cluster DB 133, and the universal cluster extraction unit 103c reads, for example, “from the universal cluster DB 133 according to the designated analysis target period. “UC-1” to “UC-3” are extracted.

  First, the similarity determination processing unit 103a compares the individual company clusters “X-1” to “X-3” with the universal clusters “UC-1” to “UC-3” extracted by the universal cluster extraction unit 103c. (Step St41). At this time, the similarity determination processing unit 103a performs all combinations of the individual company clusters “X-1” to “X-3” and the universal clusters “UC-1” to “UC-3” as indicated by reference numeral C11. Whether or not there is similarity is determined. The similarity determination means is as described with reference to FIG.

  Next, as a result of the comparison, if the individual company cluster is not similar to any of the extracted universal clusters (No in step St42), the cluster processing unit 103 sets the individual company cluster as a characteristic customer group and performs an analysis result. The data is stored in the DB 135 (step St43), and the process ends. In other words, the cluster processing unit 103 generates an integrated customer group (universal cluster) based on the purchase history data 230 that is generated based on the purchase history data 230 of the analysis target department store X and is included in the specified analysis target period. A customer group (individual company cluster) that does not have a predetermined similarity relationship is detected.

  If the individual company cluster is similar to any of the extracted universal clusters (Yes in step St42), the cluster processing unit 103 detects a difference between the individual company cluster and the universal cluster (step St48), and performs processing. Exit. Information indicating the difference between the individual company cluster and the universal cluster is stored in the analysis result DB 135. Here, the information indicating the difference is as described with reference to FIG. In this way, the cluster comparison analysis process is performed.

  As described above, the cluster processing unit 103 includes an integrated customer group (universal cluster) including all customer groups having a predetermined similarity relationship, and a customer group (based on the purchase history data 230 of the department store X to be analyzed) (Comparison cluster) is output. As described above, the comparison result (customer analysis result) is temporarily stored in the analysis result DB 135.

  Referring to FIG. 22 again, the cluster processing unit 103 reads the analysis results (comparison results) obtained in the above-described steps St43 and St48 from the analysis result DB 135 and outputs them to the terminal device 2 via the communication processing unit 14 ( Send) (step St36). In the department stores A to X, the analysis result is taken out from the terminal device 2 and used for planning a business strategy.

  Next, the cluster processing unit 103 executes a universal cluster construction process (step St37). The universal cluster construction process is as described with reference to FIG. 15B, and at least one of the universal cluster DB 133 and the universal cluster candidate DB 134 is updated by this process. After the universal cluster construction process, the customer analysis process ends.

  As described above, the analysis result is used for planning a business strategy in the department store X to be analyzed. Below, the customer analysis example 2 is given and demonstrated about utilization of an analysis result.

(Customer analysis example 2)
In this example, the cluster before the customer analysis processing is the same as the cluster of the customer analysis example 1 shown in FIG. 18, and the individual company cluster of the department store X to be analyzed is as shown in FIG. “X-1 (2012)” to “X-5 (2012)”. Further, it is assumed that the purchase history data 230 to be analyzed is for one year in 2012, and a period after 2011 is designated as an analysis target period for a short-term trend comparison analysis.

  Therefore, the universal cluster extraction unit 103c includes, from the universal cluster DB 133, “UC-2”, “UC-3”, “UC-8” including only individual company clusters whose “generation year” information indicates “2011” or later. , And “UC-10” are extracted (the process of step St34 in FIG. 22). In addition, the similarity determination processing unit 103a includes the individual company clusters “X-1 (2012)” to “X-5 (2012)” of the department store X to be analyzed, the extracted “UC-2”, “UC-3”. ”,“ UC-8 ”, and“ UC-10 ”(step St41 in FIG. 24).

  FIG. 25 shows the result of the cluster comparison analysis process in the customer analysis example 2. In the comparative analysis process, “X-1 (2012)” to “X-5 (2012)” are compared with “UC-2”, “UC-3”, “UC-8”, and “UC-10”. As a result, it is determined that “X-1 (2012)” has no similarity to any universal cluster, and thus is output as a characteristic customer group.

  The characteristic customer group “X-1 (2012)” is analyzed as a unique customer group that does not exist in the other department stores A, B, C,... It will be used to examine measures to increase sales while maintaining sales. Such measures include, for example, enriching branded products that the customer group frequently purchases (product measures).

  On the other hand, since it is determined that “X-2 (2012)” to “X-5 (2012)” have similarity to any of the universal clusters, the mutual difference is output.

  FIG. 26 shows individual company clusters and universal clusters having similarities in the customer analysis example 2. In FIG. 26, individual company clusters and universal clusters similar to each other are shown in correspondence with each other by arrows. For example, “X-3 (2012)” and “X-5 (2012)” are similar to “UC-2” and “UC-3”, respectively.

  In the comparative analysis process, the individual company clusters and the universal clusters having similarities are compared with each other as described with reference to step St48 in FIG. 24, and the difference is output as a comparison result (analysis result). Here, the information indicating the difference is as described with reference to FIG. As described above, in the department store X, the information indicating the difference regarding “X-2 (2012)” to “X-5 (2012)” is, for example, an evaluation of the strength of the customer base of the department store X (so-called benchmark evaluation). Be utilized. Thereafter, the universal cluster construction process (the process of step St37) is executed, whereby the universal cluster and the universal cluster candidate are changed.

  FIG. 27 shows the universal cluster and universal cluster candidate change processing in the customer analysis example 2. More specifically, FIG. 27A shows a universal cluster update process (see step St19 in FIG. 15B). Since the universal cluster and the individual company cluster surrounded by the dotted line frame are similar to each other, they are updated to the universal cluster indicated by the arrow by being integrated. This example is based on the assumption that the similarity relationship shown in FIG. 26 is detected in the universal cluster construction process.

  In the universal cluster construction process, for example, “UC-2” and “X-3 (2012)” are integrated into “UC-2 ′”, and “UC-3” and “X-5 (2012)” are , “UC-3 ′”. At this time, “X-3 (2012)” is added to the “original cluster” information of “UC-2 ′” after the update, and the “original cluster” information of “UC-3 ′” after the update is added to , “X-5 (2012)” is added.

  FIG. 27B shows universal cluster addition processing and universal cluster candidate update processing (see steps St17 and St16 in FIG. 15B). As described with reference to step St14 in FIG. 15B, “X-1 (2012)” that is not similar to any universal cluster is a universal cluster candidate “A-4 (2010)”, “C-5 (2012)”. ) ”Is determined.

  Since “A-4 (2010)” and “X-1 (2012)” are similar to each other, they are integrated to generate a new universal cluster “UC-11”. At this time, the “original cluster” information of “UC-11” indicates “A-4 (2010)” and “X-1 (2012)”. The universal cluster candidate DB 134 stores only “C-5 (2012)”.

  FIG. 28 shows a cluster after the customer analysis processing in the customer analysis example 2. By the universal cluster construction processing, the updated universal clusters “UC-2 ′”, “UC-3 ′”, “UC-8 ′”, “UC-10 ′”, and the like are added to the universal cluster DB 133. The universal cluster “UC-11” is stored. The universal cluster candidate DB 134 stores the universal cluster candidate “C-5” that remains after the analysis processing. The universal cluster candidate “A-4” is integrated with the private company cluster “X-1” and changed to the universal cluster “UC-11”, and thus is deleted from the universal cluster candidate DB 134.

  The individual company cluster DB 132 includes individual company clusters “A-1” to “A-7” of department store A, individual company clusters “B-1” to “B-5” of department store B, and individual company clusters of department store C. “C-1” to “C-10” are stored. Furthermore, individual company clusters “X-1” to “X-12” of the department store X to be analyzed are added and stored in the individual company cluster DB 132.

  In this way, the universal cluster DB 133 is updated each time customer analysis is performed and used for further customer analysis. Therefore, as customer analysis is performed, the contents of the universal cluster DB 133 are enriched, and highly accurate customer analysis is possible.

As described above, the customer analysis program according to the second embodiment causes the computer (CPU) 10 to execute the following processing.
Process (1): An instruction to designate the period as an analysis target period, an instruction to set a predetermined period from the current or past time point as an analysis target period, and purchase history data 230 of the analysis target store are received.
Process (2): A predetermined similarity relationship among customer groups (individual company clusters) generated based on purchase history data 230 of a plurality of stores (department stores) A, B,... Each customer group included in the integrated customer group (universal cluster) including all the customer groups having the information is stored in association with the time ("generation year" information) indicated by the purchase history data 230 based on each customer group. With reference to the storage unit (HDD) 13, the integrated customer group (universal cluster) based on the purchase history data of the period included in the designated analysis target period is extracted from the integrated customer group.
Process (3): A comparison result between the extracted integrated customer group and the customer group (individual company cluster) generated based on the purchase history data 230 of the store (department store) X to be analyzed is output.

  According to the customer analysis program according to the embodiment, the comparison result obtained by comparing the customer group of the analysis target store X with the integrated customer group to which a similar customer group in the specified analysis target period belongs is output. It is possible to analyze customers taking into account the trend of the season.

In the customer analysis method according to the second embodiment, the computer (CPU) 10 executes the following steps.
Step (1): An instruction to specify the period as an analysis target period, or an instruction to set a predetermined period from the current or past time point as an analysis target period, and purchase history data 230 of the analysis target store are received.
Step (2): a predetermined similarity relationship among customer groups (individual company clusters) generated based on purchase history data 230 of a plurality of stores (department stores) A, B,... Each customer group included in the integrated customer group (universal cluster) including all the customer groups having the information is stored in association with the time ("generation year" information) indicated by the purchase history data 230 based on each customer group. With reference to the storage unit (HDD) 13, the integrated customer group (universal cluster) based on the purchase history data of the period included in the designated analysis target period is extracted from the integrated customer group.
Step (3): A comparison result between the extracted integrated customer group and the customer group (individual company cluster) generated based on the purchase history data 230 of the analysis target store (department store) X is output.

  Since the customer analysis method according to the second embodiment includes the same configuration as that of the customer analysis program according to the second embodiment, the same effects as those described above can be obtained.

  The customer analysis apparatus 1 according to the second embodiment includes a reception processing unit (data input processing unit) 101, a storage unit (HDD) 13, and an output processing unit (cluster processing unit) 103. The reception processing unit 101 receives an instruction to designate a period as an analysis target period, or an instruction to set a predetermined period from the current or past time point as an analysis target period, and purchase history data 230 of the analysis target store.

  The storage unit 13 has a predetermined similarity relationship among customer groups (individual company clusters) generated based on purchase history data 230 of a plurality of stores (department stores) A, B,... Each customer group included in the integrated customer group (universal cluster) including all the customer groups having the information is stored in association with the time ("generation year" information) indicated by the purchase history data 230 based on each customer group. To do.

  The output processing unit 103 refers to the storage unit 13 and extracts an integrated customer group (universal cluster) based on the purchase history data 230 of the period included in the instructed analysis target period from the integrated customer group. The output processing unit 103 outputs a comparison result between the extracted integrated customer group and the customer group (individual company cluster) generated based on the purchase history data 230 of the analysis target store (department store) X.

  Since the customer analysis device 1 according to the second embodiment includes the same configuration as the customer analysis program according to the second embodiment, the same effects as those described above are achieved.

  The above processing functions can be realized by a computer. In that case, a program describing the processing contents of the functions that the processing apparatus should have is provided. By executing the program on a computer, the above processing functions are realized on the computer. The program describing the processing contents can be recorded on a computer-readable recording medium (except for a carrier wave).

  The computer that executes the program stores, for example, the program recorded on the portable recording medium or the program transferred from the server computer in its own storage device. Then, the computer reads the program from its own storage device and executes processing according to the program. The computer can also read the program directly from the portable recording medium and execute processing according to the program. Further, each time the program is transferred from the server computer, the computer can sequentially execute processing according to the received program.

  The above-described embodiment is an example of a preferred embodiment of the present invention. However, the present invention is not limited to this, and various modifications can be made without departing from the scope of the present invention.

1 customer analysis device 2 terminal device 10 CPU
13 HDD (storage unit)
101 Data input processing unit (acceptance processing unit)
102 cluster generation unit 103 cluster processing unit (output processing unit)
132 Company Cluster Database 133 Universal Cluster Database

Claims (6)

Accepts an instruction to specify the period as an analysis period, or an instruction to set a predetermined period from the current or past time point as an analysis period, and purchase history data of the store to be analyzed,
An integrated customer group including all customer groups having a predetermined similar relationship among customer groups generated based on purchase history data of a plurality of stores in the specified analysis target period, and purchase of the analysis target store A customer analysis program characterized by causing a computer to execute a process for outputting a comparison result with a customer group generated based on historical data. Accepts an instruction to specify the period as an analysis period, or an instruction to set a predetermined period from the current or past time point as an analysis period, and purchase history data of the store to be analyzed,
Each customer group included in an integrated customer group including all customer groups having a predetermined similar relationship among customer groups generated based on purchase history data of a plurality of stores in the specified analysis target period, With reference to the storage unit stored in association with the time indicated by the purchase history data based on the customer group, the purchase history data at the time included in the instructed analysis target period in the integrated customer group The integrated customer group based on
A customer analysis program that causes a computer to execute a process of outputting a comparison result between the extracted integrated customer group and a customer group generated based on purchase history data of the analysis target store. An instruction to designate the period as an analysis target period, or an instruction to set a predetermined period from the current or past time point as an analysis target period, and purchase history data of the analysis target store;
An integrated customer group including all customer groups having a predetermined similar relationship among customer groups generated based on purchase history data of a plurality of stores in the specified analysis target period, and purchase of the analysis target store A customer analysis method characterized in that a computer executes a step of outputting a comparison result with a customer group generated based on historical data. An instruction to designate the period as an analysis target period, or an instruction to set a predetermined period from the current or past time point as an analysis target period, and purchase history data of the analysis target store;
Each customer group included in an integrated customer group including all customer groups having a predetermined similar relationship among customer groups generated based on purchase history data of a plurality of stores in the specified analysis target period, With reference to the storage unit stored in association with the time indicated by the purchase history data based on the customer group, the purchase history data at the time included in the instructed analysis target period in the integrated customer group A computer that extracts the integrated customer group based on the database, and outputs the comparison result between the extracted integrated customer group and the customer group generated based on the purchase history data of the analysis target store. A customer analysis method characterized by performing. An acceptance processing unit that receives an instruction to designate a period as an analysis target period, or an instruction to set a predetermined period from the current or past time point as an analysis target period, and purchase history data of a store to be analyzed;
An integrated customer group including all customer groups having a predetermined similar relationship among customer groups generated based on purchase history data of a plurality of stores in the specified analysis target period, and purchase of the analysis target store An output processing unit that outputs a comparison result with a customer group generated based on history data. An acceptance processing unit that receives an instruction to designate a period as an analysis target period, or an instruction to set a predetermined period from the current or past time point as an analysis target period, and purchase history data of a store to be analyzed;
Each customer group included in an integrated customer group including all customer groups having a predetermined similar relationship among customer groups generated based on purchase history data of a plurality of stores in the specified analysis target period, A storage unit stored in association with the time indicated by the purchase history data based on the customer group;
Referring to the storage unit, extract the integrated customer group based on the purchase history data at the time included in the instructed analysis target period from the integrated customer group, and extract the integrated customer group, An output processing unit for outputting a comparison result with a customer group generated based on purchase history data of the analysis target store.

Images