JP2014241061A - Customer information analysis system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform clustering of each area into an area group (cluster) having a common feature based on an analysis index.SOLUTION: An existing customer database 11 accumulates existing customer information including area data which indicates an area where plural existing customers reside and plural pieces of customer relation data (saving amount and the like) which are indicated by a numeric value for every existing customer. A local information database 12 accumulates local information including the area data and plural pieces of area relation data (population density and the like) which are indicated by a numeric value for every area. A classification shaft extraction part 141 extracts a classification shaft having a predetermined correlation with respect to the analysis index selected from the respective customer relation data, from the respective customer relation data and the respective area relation data. A cluster structure analysis part 143 acquires a numeric range of the classification shaft in the respective cluster. A clustering execution part 144 performs clustering of the area into one of the respective clusters based on the numeric range of the classification shaft of the respective cluster and the numeric value of the classification shaft of the respective area.

Description

  The present invention relates to a customer information analysis system that analyzes customer information and provides support information for sales activities and marketing activities.

  It is natural to deepen business relationships with customers and increase customer loyalty in corporate sales and marketing activities. For this purpose, information related to the region (hereinafter referred to as “area information” in some cases) is analyzed, and the region's “unique” It is useful to identify “sex” and “heterogeneity”. For this reason, there is a need for an analysis technique for clarifying the characteristics of each area from the demographic information of the area information and the consumption purchasing power information for the purpose of supporting sales and marketing activities.

  On the other hand, there is a technology for analyzing area information for the purpose of supporting sales activities and marketing activities. For example, Patent Literature 1 discloses a sales information analysis apparatus that calculates an increase rate of a product sales amount for each area cluster that is a group of areas having similar attributes related to residents for the purpose of evaluating sales promotion measures and the like. In addition, Patent Document 2 aims to select a promising customer from a large number of customers. For the same area cluster as described in Patent Document 1, the area cluster is determined based on the revenue obtained from the customer. Disclosed is an information analysis device that identifies and identifies promising customers based on the revenue of each customer who resides in the identified area cluster.

JP 2009-169698 A JP 2011-211883 A

However, the devices described in Patent Document 1 and Patent Document 2 are used to evaluate sales promotion measures and identify promising customers for each predetermined area cluster, and extract the characteristics of each area. is not.
By the way, in the analysis for the purpose of supporting general sales activities and marketing activities, it is possible to plan an accurate and precise strategy by providing an analysis result based on an analysis index from many viewpoints. Desirable analysis indices are several to several tens of kinds.

  An object of the present invention is to provide a customer information analysis system for clustering each area into an area group (cluster) having a common feature based on a selected analysis index and clarifying the feature of each area.

In order to achieve the above object, the customer information analysis system of the present invention provides:
An existing customer database for accumulating existing customer information including area data indicating an area where a plurality of existing customers reside, and a plurality of customer-related data indicated numerically for each existing customer;
A regional information database that accumulates regional information including area data indicating the area, and a plurality of area-related data indicated numerically for each area;
A classification axis extracting means for extracting a classification axis having a predetermined relevance with respect to an analysis index selected from the customer related data from the customer related data and the area related data;
The analysis index is divided into a predetermined number of numerical ranges, a group of areas corresponding to each numerical range is extracted, and each cluster is based on the numerical value of the classification axis in each area included in each extracted group. Cluster structure analysis means for obtaining the numerical range of the classification axis,
Clustering means for clustering each area into one of the clusters based on the numerical range of the classification axis in each cluster determined by the cluster structure analysis means, and the numerical value of the classification axis in each area;
It is characterized by providing.

Preferably, the customer information analysis system of the present invention is
The numerical value of the analysis index in each area clustered into each cluster by the clustering means is compared with the numerical range of the analysis index in each clustered cluster, and the analysis index in the cluster in which the numerical value of the analysis index in each area is clustered An inter-area comparison means for extracting an area not included in the numerical value range is provided.

Preferably, the customer information analysis system of the present invention is
The existing customer database accumulates existing customer information collected at a plurality of points in time,
The regional information regional information database accumulates regional information collected at a plurality of points in time,
Based on the existing customer information collected at a plurality of time points accumulated in the existing customer database and the regional information collected at a plurality of time points accumulated in the regional information database, Analytical data calculation means to calculate the numerical value of the analysis index and the classification axis in the area,
The cluster structure analyzing means, the clustering means, and the inter-area comparing means, each area at a predetermined future time point determined by the analysis data calculating means as an analysis index value and a classification axis value in each area. Using the numerical value of the analysis index and the numerical value of the classification axis in
It is characterized by that.

Preferably, the customer information analysis system of the present invention is
A product share for each area can be used as the analysis index.

  According to the present invention, each area can be clustered into an area group (cluster) having a common feature based on the selected analysis index, and the feature of each area can be clarified.

It is a figure which shows an example of a structure of the customer information analysis system which concerns on embodiment of this invention. It is a figure which shows an example of the existing customer information accumulate | stored in the existing customer database. It is a figure which shows an example of the regional information accumulate | stored in the regional information database. It is a figure which shows an example of the input screen displayed on an input / output device. It is a figure which shows an example of the output screen displayed on an input / output device. It is a flowchart which shows an example of the flow of the classification axis extraction process in a classification axis extraction part. It is a figure which shows the image of an example of a classification axis extraction process. It is a flowchart which shows an example of the flow of the analysis data calculation process in an analysis data calculation part. It is a figure which shows the image of an example of an analysis data calculation process. It is a flowchart which shows an example of the regional cluster structure analysis process in a regional cluster structure analysis part. It is a figure which shows the image of an example of a regional cluster structure analysis process. It is a flowchart which shows an example of the clustering process in a clustering implementation part. It is a figure which shows the image of an example of a clustering process. It is a flowchart which shows an example of the comparison process between areas in the comparison part between areas. It is a figure which shows the image of an example of the comparison process between areas.

  Hereinafter, a customer information analysis system according to an embodiment of the present invention will be described with reference to the drawings. In all the drawings for explaining the embodiments, common constituent elements are denoted by the same reference numerals, and repeated explanation is omitted.

FIG. 1 shows an example of the configuration of a customer information analysis system 1 according to an embodiment of the present invention.
The customer information analysis system 1 includes an existing customer database 11, a regional information database 12, an input / output device 13, and a customer information analysis device 14.
The customer information analysis device 14 includes a CPU (Central Processing Unit), a main memory composed of a RAM (Random Access Memory) and the like, and a storage device composed of a hard disk and the like. The storage device of the customer information analysis device 14 stores a customer information analysis program. The CPU of the customer information analysis device 14 reads the customer information analysis program from the storage device into the main memory and executes it, whereby the classification axis extraction unit 141, the analysis data calculation unit 142, the regional cluster structure analysis unit 143, and clustering are performed. The functions of the units 144 and the inter-area comparison unit 145 are realized.
The existing customer database 11 and the regional information database 12 are communicably connected to the customer information analysis device 14 via a network. However, the existing customer database 11, the regional information database 12, and the database management program are stored in the storage device of the customer information analysis device 14, and the CPU of the customer information analysis device 14 reads the database management program from the storage device to the main memory. The existing customer database 11 and the regional information database 12 may be accessed by executing the above.
The input / output device 13 is an information processing device such as a computer that is communicably connected to the customer information analysis device 14 via a network. However, the input / output device 13 may be directly connected to the customer information analysis device 14 and may be a keyboard and a display.

FIG. 2 shows an example of the existing customer information 21 stored in the existing customer database 11.
In the existing customer database 11, existing customer information 21 of each existing customer collected at a plurality of points in time is accumulated in chronological order. The existing customer information 21 is collected by a questionnaire survey or the like.
The existing customer information 21 includes a plurality of data associated with the customer, such as customer name, area name, age, occupation, family structure, and financial asset information such as savings, insurance premiums and purchase amount of securities.
In the following, among the data included in the existing customer information 21, what is indicated by numerical values (age, occupation, family structure, financial asset information such as savings, insurance claims and securities purchases, etc.) It is called data. The area name is an example of area data of the present invention.

FIG. 3 shows an example of the area information 31 stored in the area information database 12.
In the regional information database 12, regional information 31 of each area collected at a plurality of points in time is accumulated in chronological order.
Regional information 31 includes area name, age-specific population, financial asset information such as average savings, geographical statistical information such as population density, number of households such as the number of households per household, and number of workers in tertiary industries A plurality of data associated with such areas. The regional information 31 is collected from, for example, statistical information released by the country or local government.
In the following, among the data included in the regional information 31, those indicated by numerical values (such as population by age, financial asset information such as average savings, geographical statistical information such as population density, number of households per household member, etc.) The number of households and the number of workers in the third industry) are called area-related data.

FIG. 4 shows an example of the input screen 41 displayed on the input / output device 13.
The input screen 41 displays a list of analysis indices. The user selects an analysis index by checking a check box corresponding to the analysis index. In the example of FIG. 4, savings amount is selected as the analysis index.
As the analysis index, customer-related data, product share, and the like are used. Data that is not included in the existing customer information 21 such as product share is analyzed when data for each area collected by market research of the product is stored in the storage device of the customer information analysis device 14 or the like. It can be used as an indicator.

FIG. 5 shows an example of the output screen 51 displayed on the input / output device 13.
For example, the customer information analyzer 14 clusters each area according to the level of potential savings using the savings as an analysis index. And as shown in FIG. 5, the customer information analysis apparatus 14 displays on the map the area where the actual saving amount is low although the potential saving amount is high.

FIG. 6 is a flowchart showing an example of the flow of the classification axis extraction process in the classification axis extraction unit 141. FIG. 7 shows an example of the classification axis extraction process.
The classification axis extraction unit 141 extracts a classification axis having a predetermined relevance with respect to the analysis index (for example, savings amount) selected on the input screen 41 from the customer related data and the area related data. For example, when the saving amount selected as the analysis index is selected, the classification axis extraction unit 141 obtains a classification axis for classifying into an area where the value of the saving amount is potentially high and low.
First, the classification axis extraction unit 141 receives an analysis index designated by the user from the input screen 41 of the input / output device 13 (step 601). In FIG. 7, it is assumed that [savings amount] is designated as the analysis index.
Next, when [savings amount] is received as an analysis index, since the average savings amount is accumulated in the regional information database 12, the classification axis extraction unit 141 has the latest collection date from the regional information database 12. Extract the average savings. When data that is not stored in the regional information database 12 (for example, insurance premiums or purchases of securities) is received as an analysis index, the classification axis extraction unit 141 has the latest collection date from the existing customer database 11. And an average value is obtained for each existing customer information 21 having the same area name. Hereinafter, data with the newest collection date is referred to as existing data, and data older than the existing data is referred to as past data.
Then, the classification axis extraction unit 141 extracts existing data of customer related data and area related data from the existing customer database 11 and the regional information database 12 (step 602).

Next, the classification axis extraction unit 141 is an index indicating the relationship between the existing data of the analysis index and the data used as the determination material for determining the classification axis for the existing data of the customer-related data and the area-related data. Is calculated (step 603). As an index indicating the relationship, a correlation coefficient indicating a correlation between data and a principal component load used in principal component analysis are calculated.
Next, the classification axis extraction unit 141 determines data serving as the optimal classification axis for the analysis index from the index indicating the relationship between the calculated analysis index and the classification axis candidate (step 604). For example, customer-related data and area-related data whose correlation coefficient is close to 1 and customer-related data and area-related data whose values of main component load are large are preferentially used as classification axes. By selecting the classification axis based on such criteria, the cluster can be easily found in the next regional cluster structure analysis unit 142.
The user specifies the number of data of the classification axis determined here. The classification axis extraction unit 141 selects and determines the number of classification axis candidate data in the order from the largest relationship index. In FIG. 7, the analysis index is designated as [savings amount] and the number of data on the classification axis is two, and [corresponding to [savings amount] and the correlation coefficient of each data are calculated and the top two [ The flow of determining the number of household members per household] and [population density] as the classification axes is shown.

FIG. 8 is a flowchart illustrating an example of the flow of analysis data calculation processing in the analysis data calculation unit 142. FIG. 9 shows an example of an analysis data calculation process.
Since existing customer information 21 and regional information 31 also include data that is updated every few years, information that shows the current or future environment necessary for strategic planning of sales and marketing activities even if used as is May not be available. Therefore, the analysis data calculation unit 142 has a predetermined time point in consideration of future growth potential based on the time series data of the analysis index received in step 601 and the time series data of the classification axis determined in step 604. Estimate (determine) the numerical value of the analysis index and the numerical value of the classification axis in each area.
First, the analysis data calculation unit 142 acquires the past data and existing data of the analysis index and the data serving as the classification axis from the existing customer database 11 and the regional information database 12, and acquires time-series data (step 801). FIG. 9 shows an example of acquiring past data and existing data of [number of household members per household] from the regional information database 12.
Next, regression analysis is performed using the time series data acquired in step 801 as an objective variable and the time axis as an explanatory variable, and a regression equation is calculated (step 802). In FIG. 9, the vertical axis indicates [number of household members per household] and the horizontal axis indicates the time axis (2010/4 to 2013/4). Regression formula [number of household members per household] = α × [time axis ] An example of calculating + β is shown.
Next, using the regression equation calculated in step 802, the value of the objective variable considering the growth potential after a certain period is estimated (step 803). In FIG. 9, the regression formula [number of household members per household] calculated in step 802 = [alpha] × [time axis] + [time axis] = 2014/4 is substituted for [household personnel at 2014/4]. The number of households per person] is estimated to be 142 cases.

FIG. 10 is a flowchart illustrating an example of the regional cluster structure analysis process in the regional cluster structure analysis unit 143. FIG. 11 shows an example of the regional cluster structure analysis process.
The regional cluster structure analysis unit 143 obtains a cluster into which each area is classified from the analysis index obtained by the analysis data calculation unit 142 in consideration of the growth and the data of the classification axis. Specifically, as will be described below, the regional cluster structure analysis unit 143 divides the analysis index into a predetermined number of numerical ranges, extracts (determines) groups of areas corresponding to the respective numerical ranges, and extracts them. The numerical range of the classification axis in each cluster is obtained based on the numerical value of the classification axis in each area included in each group.
First, the regional cluster structure analysis unit 143 receives the analysis index and the classification axis data in consideration of the growth potential calculated by the analysis data extraction unit 142 (step 1001). The data shown on the left side of FIG. 9 receives growth-related data for [Number of households by household members] and [Population density], and [Number of households by household members] on the X axis and [Population density] on the Y axis. Each area is arranged by designating.
Next, when analyzing the cluster structure for the data that takes into consideration the growth potential received in step 1001, the regional cluster structure analysis unit 143 filters the area group for the purpose of facilitating the analysis. This filter content is, for example, that each area is sorted by analysis index data, and only the areas of the upper area group whose value is included in the upper 10% and the lower area group included in the lower 10% are analyzed. Yes (step 1002). This is because if all areas are analyzed, the number of areas is large and the boundaries between clusters become unclear, making it difficult to determine the cluster structure. For this reason, the analysis of the cluster structure is facilitated by focusing on only the upper area group and the lower area group of the analysis index that seems to have the most difference in the characteristics between the areas. In FIG. 11, it is difficult to analyze the cluster structure in the scatter diagram in which all the areas on the lower left side are arranged, but in the scatter diagram on the middle lower side after filtering, it can be seen that clusters exist in the upper right and lower left.
Next, the regional cluster structure analysis unit 143 determines a cluster structure for the upper area group and the lower area group of the analysis target area data calculated in Step 1002 (Step 1003). FIG. 11 shows an example in which the upper right cluster is determined as the upper savings cluster and the two lower left clusters are determined as the lower savings clusters.

FIG. 12 is a flowchart illustrating an example of clustering processing in the clustering execution unit 144. FIG. 13 shows an example of clustering processing.
The clustering execution unit 144 assigns all areas to the clusters determined by the regional cluster structure analysis unit 143. Specifically, as will be described below, the clustering execution unit 144 sets each classification axis in each cluster obtained by the cluster structure analysis unit 143 and the numerical value of the classification axis in each area. Cluster the area into one of each cluster.
First, the clustering execution unit 144 calculates the distances between all the clusters determined by the regional cluster structure analysis unit 143 and all the areas (Step 1201). The distance here refers to the similarity between the features of the cluster and the features of the target area. Therefore, the Mahalanobis distance is calculated based on the average attribute value of each area constituting the cluster and the attribute value of the target area.
Next, based on the distance calculated in step 1201, it is determined that the cluster at the shortest distance from the target area is the cluster having the highest similarity, and is determined as the cluster belonging to the target area (step 1202). ).
The example of FIG. 13 shows an image when the analysis index is [savings] and the classification axis is [number of households per household] and [population density]. The distance between each cluster and all areas determined by the regional cluster structure analysis unit 143 is calculated and assigned to each cluster.

FIG. 14 is a flowchart illustrating an example of an inter-area comparison process in the inter-area comparison unit 145. FIG. 15 shows an example of an inter-area comparison process.
As will be described below, the inter-area comparison unit 145 compares the numerical value of the analysis index in each area clustered into each cluster by the clustering execution unit 144 and the numerical range of the analysis index in the clustered cluster. An area that is not included in the numerical range of the analysis index in the cluster in which the numerical value of the analysis index in the area is clustered is extracted.
First, the inter-area comparison unit 145 calculates a potential class (value) of the analysis index from the cluster to which each area obtained by the clustering execution unit 144 is obtained. Further, the inter-area comparison unit 145 calculates an actual measurement class (value) from the actual analysis index of each area (step 1401).
Next, the inter-area comparison unit 145 displays areas where the calculated potential class of the analysis index is different from the actually measured class (step 1402).
FIG. 15 shows an example in which, for example, an area with an average saving amount of 15 million yen or more is an upper area, and an area with an average saving amount of less than 15 million yen is a lower area.
The inter-area comparison unit 145 calculates the estimated class calculated as the upper area and the lower area of the [savings amount] from the cluster obtained by the clustering execution unit 144 and the upper area and the lower area from the measured value of the [savings amount]. By comparing the classes, an area where the savings amount is low (the measured class is the lower area) although the [saving amount] is potentially high (the estimated class is the upper area) is specified, and the area is output to the output screen 51. To display. The displayed area has a feature that the potential savings are higher than the actual savings, so that many customers whose savings will increase in the future reside. For this reason, it is considered that the sales promotion effect is enhanced by setting the area as a priority area of sales activities.

In the above-described embodiment, the classification axis candidates are the customer-related data and the area-related data, but in addition to this, the product share and the like may be the classification axis candidates. However, for data not included in the existing customer information 21 and regional information 31 such as product share, data for each area collected by market research of the product is stored in the storage device of the customer information analysis device 14 or the like. Can be a candidate for a classification axis.
In the above-described embodiment, the regional cluster structure analysis unit 143, the clustering execution unit 144, and the inter-area comparison unit 145 are obtained by the analysis data calculation unit 142 as the numerical value of the analysis index and the numerical value of the classification axis in each area. Although it is assumed that data considering growth is used, each of these units may use existing data as the numerical value of the analysis index and the numerical value of the classification axis in each area.
In the above-described embodiment, an example is shown in which each area is clustered into two clusters, the upper cluster and the lower cluster, of the analysis index (saving amount). However, the analysis index value is higher, middle, and lower. For example, clustering may be performed on three or more clusters. For example, you may cluster into four clusters whose potential savings are less than 10 million yen, 10 million yen-15 million yen, 15 million yen-20 million yen, and 20 million yen or more.

As described above, according to the present invention, in sales activity / marketing activity support using area information, the latent numerical value of the analysis index is obtained through clustering of each area based on the analysis index specified by the user, and the actual numerical value is obtained. Information that supports strategic planning and decision making can be provided.
Furthermore, it is possible to provide information that supports strategy planning corresponding to changes by clustering each area using data that takes into account the growth potential of each area.

DESCRIPTION OF SYMBOLS 1 ... Customer information analysis system 11 ... Existing customer database 12 ... Area information database 13 ... Input / output device 14 ... Customer information analysis device 141 ... Classification axis extraction ancestor part 142 ... Analysis data calculation unit 143 ... regional cluster structure analysis unit 144 ... clustering execution unit 145 ... inter-area comparison unit 21 ... existing customer information 31 ... regional information 41 ... input screen example 51 ..Output screen examples

Claims (4)

An existing customer database for accumulating existing customer information including area data indicating an area where a plurality of existing customers reside, and a plurality of customer-related data indicated numerically for each existing customer;
A regional information database that accumulates regional information including area data indicating the area, and a plurality of area-related data indicated numerically for each area;
A classification axis extracting means for extracting a classification axis having a predetermined relevance with respect to an analysis index selected from the customer related data from the customer related data and the area related data;
The analysis index is divided into a predetermined number of numerical ranges, a group of areas corresponding to each numerical range is extracted, and each cluster is based on the numerical value of the classification axis in each area included in each extracted group. Cluster structure analysis means for obtaining the numerical range of the classification axis,
Clustering means for clustering each area into one of the clusters based on the numerical range of the classification axis in each cluster determined by the cluster structure analysis means, and the numerical value of the classification axis in each area;
A customer information analysis system characterized by comprising:   The numerical value of the analysis indicator in each area clustered into each cluster by the clustering means is compared with the numerical range of the analysis indicator in the clustered cluster, and the analysis indicator in the cluster in which the numerical value of the analysis indicator in each area is clustered The customer information analysis system according to claim 1, further comprising an inter-area comparison unit that extracts an area that is not included in the numerical range. The existing customer database accumulates existing customer information collected at a plurality of points in time,
The regional information regional information database accumulates regional information collected at a plurality of points in time,
Based on the existing customer information collected at a plurality of time points accumulated in the existing customer database and the regional information collected at a plurality of time points accumulated in the regional information database, Analytical data calculation means to calculate the numerical value of the analysis index and the classification axis in the area,
The cluster structure analyzing means, the clustering means, and the inter-area comparing means, each area at a predetermined future time point determined by the analysis data calculating means as an analysis index value and a classification axis value in each area. Using the numerical value of the analysis index and the numerical value of the classification axis in
The customer information analysis system according to claim 2, wherein:   The customer information analysis system according to any one of claims 1 to 3, wherein a product share for each area can be used as the analysis index.

Images