JP2018067112A - Salesperson clustering system and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a salesperson clustering system capable of generating cluster variables which eliminate arbitrariness and effectively capture characteristics of salespersons.SOLUTION: A salesperson clustering system 10 comprises: salesperson data storage means 30 for storing the amount of profit of each customer for each salesperson in association with salesperson identification information and customer identification information; feature vector creating means 22 for calculating respectively the amount of profit at a plurality of quantiles in profit distribution for each salesperson using stored data on the amount of profit, and creating feature vectors for each salesperson which includes each of the calculated amount of profit as constituent elements, where the cluster variables for a vertical axis of the profit distribution is the number of customers, the cluster variables for a horizontal axis of profit distribution is the amount of the profit from one customer; and cluster analyzing means 23 for classifying the plurality of salespersons according to cluster analysis using the created feature vectors for each salesperson as input.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a salesperson clustering system and program configured by a computer that executes processing for classifying a plurality of salespersons who make a profit by contacting customers, and belongs to an organization such as a company composed of a plurality of departments, for example. This can be used to support the task of reassigning multiple sales employees due to personnel changes.

  Conventionally, when a “person” is segmented, such as a salesperson engaged in operations related to various goods and services such as finance, insurance, and door-to-door sales of products, personal profiles such as age and gender, Create a cluster variable from the aggregated values of activities such as the number of points and sales revenue, put these variables into the clustering algorithm for cluster analysis, and group the results by calculating the distance between salespeople Many have been adopted.

  As a technique related to the present invention, a sales support system that analyzes sales activities of salespeople and proposes improvement policies is known (see Patent Document 1). In this sales support system, the sales staff can be instructed by judging that the number of visits by the sales staff is small. For example, if the number of visits and visit times within a given period (or cumulative) for a customer exceed the visit and visit thresholds described in the rule, the message “ The number of visits and time are large. Review the proposed strategy ”is displayed on the screen (see paragraph [0078], FIG. 5). Similarly, by comparison with the rules, the message “It seems that the proposal activities for specific customers are concentrated. There are few new customer visits. It seems that the visit time per customer is long. Please increase the number of visits ”is displayed on the screen (see paragraph [0085], FIG. 6).

  At this time, in the rule calculation, for example, when “the number of visits is small”, the threshold is determined based on the lower α percentage point of the distribution from the statistical distribution regarding the number of visits of all the salespeople in the area to which the salesperson belongs ( Paragraph [0093]). For example, when the percentage value representing the threshold is the lower 5%, if the number of visits of a sales employee is in the lower 5% group of the overall distribution, it is determined that the number of visits of the sales employee is small ( Paragraph [0093]). Implicit parameters are used when performing rule calculations. For example, in the case of a rule that the number of visits is 25% or less compared to the average value of surrounding sales staff, “25%” is implicit. Parameter (see paragraph [0098]). When converting the generated guidance rules into machine rules, adjectives such as “short”, “relatively long”, and “less” are expressed by threshold values using statistical information, and threshold values using this statistical information are used. Is expressed by, for example, the upper or lower percentile (percentage) of the distribution of the feature quantity, and the percentile is an example, and an arbitrary quantile (eg, quartile) is used. (See paragraph [0114]).

  Therefore, in the sales support system described in Patent Document 1, the upper or lower percentile in the distribution or an arbitrary quantile (for example, quartile) is used for the number of visits. As described above, the quantile in the distribution is the quantile in the distribution of the number of visits of all sales employees and peripheral sales employees (that is, a plurality of sales employees). According to claims 2 and 7 of patent document 1, “the threshold value of the feature amount of the business activity is the upper quantile and lower value in the distribution of the feature amount of the business activity already acquired from the plurality of sales staff”. It is clear from the description that “at least one of the side quantiles”, and the distribution of Patent Document 1 is a distribution of a plurality of sales employees. It means the number of sales employees from the top or bottom. On the other hand, as will be described later, the distribution of the present invention is the distribution of revenue for each salesperson, the distribution of the number of contacts for each salesperson, that is, the distribution of one salesperson (one salesperson is in charge). (The distribution formed by data on each of a plurality of customers), and the quantile (quantile point) in the distribution of the present invention is the number of customers from the top or the bottom of the plurality of customers. In this respect, processing completely different from the business support system described in Patent Document 1 is performed.

JP 2012-8947 A (paragraphs [0078], [0085], [0093], [0098], [0114], FIG. 5, FIG. 6, claims 2 and 7)

  The conventional method described above has the following problems. First, all the cluster variables were created by the data designer or analyst, and the clustering results were also clustered because the analyst arbitrarily extracted these cluster variables during clustering. It tends to be arbitrary reflecting the criteria for extracting variables.

  Next, when using numerical data (revenue amount, contact count) indicating salesperson's performance as a cluster variable, it is aggregated into one representative value as an aggregated value, such as the total value or average value. In many cases, information other than the representative value was lost.

  For this reason, for example, when using clustering, which is one of data mining methods, for the purpose of segmentation of sales staff engaged in operations related to various goods and services such as finance, insurance, and door-to-door sales of products, It is desirable to develop a technique for eliminating the arbitraryness from the creation of cluster variables (features) and creating feature quantities that effectively capture the features of salespeople.

  In the sales support system described in Patent Document 1 described above, sales support processing is performed according to a rule determined using the quantile in the distribution of the number of visits. Since it only shows the position of the salesperson with respect to the salesperson (that is, a plurality of salespersons), as described above, the process of creating feature quantities (feature vectors) that effectively capture the characteristics of the salesperson I am not doing.

  An object of the present invention is to provide a salesperson clustering system and program capable of creating a cluster variable that eliminates arbitraryness and effectively captures the characteristics of a salesperson.

The present invention is a salesperson clustering system configured by a computer that executes a process of classifying a plurality of salespersons who make profits in contact with customers,
Salesperson data storage means for storing the revenue amount for each salesperson and for each customer in association with the salesperson identification information and the customer identification information;
Revenue distribution for each salesperson using the data on the revenue amount stored in this salesperson data storage means, with the vertical axis representing the number of customers and the horizontal axis representing the revenue amount raised from one customer, as cluster variables. A feature vector creating means for obtaining each revenue amount of a plurality of quantiles in, and executing a process of creating a feature vector having each calculated revenue amount as a constituent element for each salesperson;
Cluster analysis means for executing a process of classifying a plurality of salespeople by cluster analysis using the feature vector for each salesperson created by the feature vector creation means; and
The results of classification by this cluster analysis means and / or personnel transfer support processing using the classification results or other processing results are displayed on the screen, printed, transmitted to other systems via a communication line, and stored in a recording medium. Or an output means for executing a process of outputting sound.

  Here, “multiple quantiles in the distribution of revenue for each salesperson” preferably includes quartiles (first, second, and third quartiles), but any quantile is included. The number of quantiles may be two, three, or four or more. In addition, the “multiple quantiles” may be distributed on the smaller side and the larger side of the amount of earnings (median) of the second quartile, You may provide only in any one side seeing from a profit amount (median value). For example, two quantiles such as an upper 5% quantile (lower 95% quantile) and an upper 40% quantile (lower 60% quantile) may be used. Furthermore, weighting may be performed by increasing the dimension of the feature vector by putting the revenue amount of the same quantile as a plurality of elements.

  In such a salesperson clustering system of the present invention, as a variable for the cluster, a feature vector having each revenue amount of a plurality of quantiles in the revenue distribution for each salesperson as a constituent element is created for each salesperson. Since the cluster analysis is performed using the feature vector as input, it is possible to eliminate arbitraryness when creating a variable for the cluster and to create a variable for the cluster that effectively captures the characteristics of the salesperson Thus, the problem of the conventional method that uses the total value such as the total value or the average value as the cluster variable is solved.

  For example, in the conventional method, an influence such as when a customer with a large profit amount can be acquired by chance (so-called lucky) during the aggregation target period may be reflected in the classification result. In the salesperson clustering system, it is possible to grasp the individual profit characteristics of the salesperson by a plurality of quantiles, so that it is less likely to be affected by accidents or unusual situations during such a counting target period.

  In other words, in the past, salesmen's quotas were often sales, and so on. In many cases, simple evaluations were made based on the sales amount. It is important that good business results are obtained), and since the present invention creates vector variables for clusters using multiple quantiles in the revenue distribution, such customer portfolio features It is possible to create a feature vector to be shown and perform stable classification. Because the conventional method makes a simple evaluation based on the sales amount, a salesperson who has one customer of 10 million yen and a salesperson who has 1000 customers of 10,000 yen They are classified into the same cluster. However, because there are years when it is not possible to acquire a customer of 10 million yen, the evaluation is not stable.

  In addition, by introducing the salesperson clustering system of the present invention, organizations such as companies with a large number of salespersons can pattern the individual earnings characteristics of each salesperson, creating an effective team organization, etc. Will be able to formulate a sales strategy.

  Furthermore, since the cluster variable is a feature vector having the quantile profit amount as a constituent element, the same clustering can be performed without changing the definition of the vector variable according to the application. The objective is achieved.

  <Details of quantile>

In the sales staff clustering system described above,
Feature vector creation means
The three quartiles, the first quartile, the second quartile, and the third quartile, and the third quartile with the largest revenue among these three quartiles It is configured to execute a process of creating for each sales employee a four-dimensional or higher feature vector having the revenue amount of at least one higher quantile subdivided in a range with a larger revenue amount as a component. Is desirable.

  In this way, when a configuration using a four-dimensional or higher feature vector having each quartile and the higher quantile as a constituent element is used, the individual income of each salesperson is determined by the quartile. Capturing the characteristics evenly (no particular lack) and creating vector variables with finer steps in the tail of the revenue distribution (the bottom of the revenue) Therefore, in accordance with Pareto's law (the majority of the overall features are the empirical rules that some of the elements that make up the whole are creating, also called the 80:20 law), the tail part It is possible to perform cluster analysis with emphasis on the above.

Furthermore, as described above, in the case of a configuration using a four-dimensional or higher feature vector having each quartile and each higher monetary amount as a component,
Feature vector creation means
Use the number of customers in charge input by the user for the range, average number, maximum number, or minimum number of customers in charge of one sales employee, or use data stored in the salesperson data storage means Calculate the number of responsible customers, and determine the number and step size of the upper quantiles to be subdivided according to the correspondence between the predetermined number of responsible customers and the number and step size of the upper quantiles. It is desirable that the processing is executed.

  Here, the “range of the number of customers” is a range such as less than 100, 100 to 199, 200 to 399, 400 to 999, 1000 or more. Further, the “correspondence relationship between the number of customers in charge and the number and step size of higher quantiles” may be described in the program or stored in a memory (quantile point storage means).

  In this way, when the correspondence relationship between the number of customers in charge, the number of higher quantiles and the step size is determined in advance, the number of upper quantiles and the step size are determined according to the correspondence relationship. In accordance with Pareto's Law, it is possible to create feature vectors according to the number of customers in charge when making finer increments in the tail part of the revenue distribution. Creation can be realized. That is, when the number of customers in charge is large, the number of higher quantiles can be increased and the step size of the higher quantiles can be made narrower (smaller). For example, the top 0.1% quantile when the number of customers in charge is 1,000 indicates the revenue amount of the first customer from the top, but when the number of customers in charge is 400, the top 0.1% Since the profit amount of the percentile may become a meaningless value, such a situation can be avoided in advance.

In the sales staff clustering system described above,
Feature vector creation means
In addition to each revenue amount of multiple quantiles, the maximum amount of revenue amount raised from each customer in charge of one sales employee and the amount of revenue amount raised from each customer in charge of one sales employee Executes the process of creating a feature vector for each salesperson that includes at least one additional element selected from the total value and the average value of the amount of revenue generated by each salesperson for each salesperson It is good also as composition to do.

  In this way, in addition to each revenue amount of multiple quantiles, if it is configured to create a feature vector with the maximum value, total value, or average value as an additional component, the total value, average value, etc. It is possible to perform cluster analysis that incorporates the concept of the conventional method that uses the aggregate value as a component of the feature vector.

  <Classification result by the first cluster analysis using a plurality of quantiles in the revenue distribution for each salesperson and classification result by a second cluster analysis using a plurality of quantiles in the distribution of the number of contacts for each salesperson Configuration for combined classification processing combining

In the sales staff clustering system described above,
Salesperson data storage means
In addition to the amount of revenue for each salesperson and for each customer, the number of contacts for each salesperson and for each customer is stored in association with the salesperson identification information and customer identification information,
Feature vector creation means
In addition to processing to create a feature vector for each salesperson with revenue amount as a component,
Using the contact count data stored in the salesperson data storage means, as a cluster variable, the vertical axis represents the number of customers and the horizontal axis represents the number of contacts for one customer. The number of contacts for each quantile is calculated, and a process for creating a feature vector for each salesperson with each calculated number of contacts as a component is also executed.
Cluster analysis means
In a first classification process for classifying a plurality of sales employees into a first divided number of clusters by a first cluster analysis using as input a feature vector for each sales employee having the revenue amount created by the feature vector creation means as a component. In addition,
A second classification process for classifying a plurality of salespersons into a second division number cluster by a second cluster analysis using a feature vector for each salesperson with the number of contacts created by the feature vector creation means as an input. Run,
Further, by combining the first classification result obtained by the first classification process and the second classification result obtained by the second classification process, a plurality of sales employees can be divided into the first division number and the first division number. It may be configured to execute a composite classification process for classifying the number of clusters by the number of divisions of 2.

  As described above, according to the classification result by the first cluster analysis using the plurality of quantiles in the revenue distribution for each salesperson and the second cluster analysis using the plurality of quantiles in the contact count distribution for each salesperson. In the case of a configuration in which the combined classification processing is combined with the classification result, it is possible to perform clustering that combines the individual profit characteristics of each salesperson and the personal service characteristics of each salesperson. For this reason, for example, it is possible to perform clustering that takes into account the personal characteristics of salespeople, such as whether they are salespeople who have a large number of contacts with specific customers or whether they are salespeople who contact all customers equally. It becomes.

  <Configuration for personnel change support processing>

In addition, in the salesperson clustering system described above (including the configuration that performs the composite classification process),
Multiple salespeople are salespeople who belong to an organization consisting of multiple departments.
Cluster analysis means
Execute at least one type of cluster analysis for all sales employees belonging to the organization, and combine each cluster obtained by one type of cluster analysis or multiple types of cluster analysis. For each cluster obtained by the combined classification process, the cluster number assigned according to the degree of excellence in sales results or the cluster feature amount indicating the degree of excellence of each cluster is used for sales for sales staff classified into each cluster. It is configured to execute processing to be stored in the classification result storage means in association with the member identification information,
With personnel change support means to execute personnel change support processing that supports the relocation work between sales staff departments,
This personnel change support means
When assigning a sales employee to each department, the cluster number or cluster feature quantity stored in the classification result storage means in association with the sales representative identification information for the transfer-side sales employee assigned to each department, and each department The cluster number or the cluster feature value stored in the classification result storage means in association with the salesperson identification information about the salesperson on the transfer side who has formed a vacancy in each department due to the transfer from one department to another Cluster number stored in the classification result storage means in association with the salesperson identification information of the two salespeople to be exchanged when performing the personnel selection process or exchanging salespeople between the two departments Alternatively, it is desirable to perform a person selection process for matching cluster feature amounts.

  Here, the “cluster feature indicating the degree of excellence of each cluster” is a representative value of the cluster that is different from the “cluster number”. For example, the feature vector of each salesperson classified into the cluster It is an average value of one of the constituent elements, and more specifically, for example, an average value of the top 0.5 revenue amounts of each salesperson. For example, in the case where the whole is divided into seven, the “cluster number” is, for example, 1, 2, 3, 4, 5, 6, 7 in order from the cluster with the lowest grade according to the degree of excellence of the business results. Although simply given, the “cluster feature” is not a linear numerical value sequence, but can be a quantity in which each cluster is weighted nonlinearly. The “cluster number” can be squared to 1, 4, 9, 16, 25, 36, 49, for example, and can be used as a “cluster feature”. A constant may be added after the square or the third power, or it may be a route. In the case of a route, the difference between the cluster numbers 1 and 7 is that the cluster feature is route 1 (= 1) and the route 7 (= 2.645575), the weighting is performed so that the difference between the two becomes small.

  In addition, “execute a person selection process for matching the cluster number or cluster feature amount” means that it is only necessary to try the matching process, and if there is no salesperson who matches the cluster number or cluster feature amount, This does not preclude the process of selecting a salesperson with a cluster number or cluster feature quantity close.

  Furthermore, “when assigning a sales employee to each department” and “when exchanging a sales person between two departments” are divided into the latter “when exchanging a sales person between two departments”. This is because not only the one-to-one trade as described above, but also the sales staff may be switched between three or more departments as in the case of the former “when assigning a sales person to each department”. Turn salespersons A, B, and C in departments α, β, and γ, that is, salesperson A in department α is transferred to department β, salesperson B in department β is transferred to department γ, and sales of department γ This is because employee C may be transferred to department α. However, the same result can be obtained for the relocation between three or more departments if the exchange between two departments is repeated a plurality of times.

  In this way, the personnel change support means, when placing a sales person in each department, execute a person selection process to match the cluster number or cluster feature amount of the sales person on the transfer side and the sales person on the transfer side, Or, when exchanging sales personnel between two departments, if the configuration is such that the selection process is performed to match the cluster numbers or cluster features of the two sales representatives to be exchanged, Before and after the transfer, sales employees classified into the same cluster can be switched, and the balance of business processing ability between departments can be maintained.

  <Configuration to perform selection processing when there are multiple candidates under the principle of assigning or exchanging salespeople classified into the same cluster in personnel transfer support processing>

Furthermore, in the salesperson clustering system described above,
Salesperson data storage means
The feature vector of each salesperson created by the feature vector creation means is also configured to be stored in association with the salesperson identification information,
Personnel transfer support means
When assigning sales employees to each department, if there are multiple sales side salespersons whose cluster numbers or cluster feature values match the cluster number or cluster feature value of the transfer side salesperson, or move out When there are multiple transfer-side salespeople whose cluster numbers or cluster feature values match the cluster number or cluster feature amount of the salesperson on the side, or when exchanging salespeople between two departments, When there are a plurality of sales personnel to be exchanged with the same cluster number or cluster feature quantity, the salesperson with the shortest distance between the two calculated using the feature vector stored in the salesperson data storage means Can be configured to execute the process of selecting

  Under the principle of arranging or exchanging salespeople classified into the same cluster in this way, when there are multiple candidates, select the salesperson with the smallest distance between the two calculated using the feature vector With this configuration, it is possible to further reduce the change in the business processing capability of each department before and after personnel changes.

  <Performance change support processing is performed, and the first and second cluster analysis is performed, but the combined classification processing is not performed by combining the classification results, but the first classification processing by the first cluster analysis is performed. The selection process when there are a plurality of candidates under the principle of using each cluster obtained in step 1 to arrange or exchange sales staff classified into the same cluster is performed as the second classification by the second cluster analysis. Configuration using each cluster obtained by processing>

And in the salesperson clustering system mentioned above,
Salesperson data storage means
In addition to the amount of revenue for each salesperson and for each customer, the number of contacts for each salesperson and for each customer is stored in association with the salesperson identification information and customer identification information,
Feature vector creation means
In addition to processing to create a feature vector for each salesperson with revenue amount as a component,
Using the contact count data stored in the salesperson data storage means, as a cluster variable, the vertical axis represents the number of customers and the horizontal axis represents the number of contacts for one customer. The number of contacts for each quantile is calculated, and a process for creating a feature vector for each salesperson with each calculated number of contacts as a component is also executed.
Multiple salespeople are salespeople who belong to an organization consisting of multiple departments.
Cluster analysis means
The first cluster analysis is used to input a feature vector for each salesperson whose component is the revenue amount created by the feature vector creation means for all salespeople belonging to the organization. In addition to the first classification process for classifying
A feature vector for each salesperson whose component is the number of contacts created by the feature vector creation means for all salespersons belonging to the organization is input as a second division number for a plurality of salespersons by the second cluster analysis. A second classification process for classifying the cluster into
Furthermore, the cluster number assigned according to the degree of excellence of the business results for each cluster obtained by the first classification process or the cluster feature amount indicating the degree of excellence of each cluster can be obtained by the first classification process. In addition, it is stored in the classification result storage means in association with the salesperson identification information about the salesperson classified into each cluster, and is attached to each cluster obtained in the second classification processing according to the degree of excellent business results. The cluster feature quantity indicating the cluster number or the degree of excellence of each cluster is stored in the classification result storage means in association with the salesperson identification information about the salesperson classified into each cluster obtained in the second classification process. With the configuration,
With personnel change support means to execute personnel change support processing that supports the relocation work between sales staff departments,
This personnel change support means
Cluster number or cluster of the first classification process stored in the classification result storage means in association with the salesperson identification information about the transfer-side salesperson arranged in each department when the salesperson is arranged in each department The first classification stored in the classification result storage means in association with the feature quantity and the salesperson identification information about the salesperson on the transfer side who has formed a vacancy in each department due to the transfer from each department to another department When a person selection process that matches the cluster number or cluster feature value of the process is executed, or when a sales employee is exchanged between two departments, it is associated with sales representative identification information about the two sales representatives to be exchanged. Performing a person selection process for matching the cluster number or cluster feature amount of the first classification process stored in the classification result storage means,
Further, when a salesperson is assigned to each department, the cluster number or cluster feature value of the first classification process matches the cluster number or cluster feature value of the first classification process for the salesperson on the transfer side. When there are a plurality of transfer side salespeople, or the cluster number or cluster feature amount of the first classification process matches the cluster number or cluster feature amount of the first classification process for the transfer side salesperson When there are multiple transfer-side salespersons or when exchanging salespersons between two departments, there are multiple sales-target salespersons with the same cluster number or cluster feature quantity in the first classification process In this case, a configuration may be adopted in which a process is performed in which the sales numbers that match or are close to the cluster number or cluster feature amount of the second classification process.

  In this way, the first and second cluster analyzes are performed. However, when the classification results are not used in combination with each other and the classification results are used step by step in different scenes. , Before and after personnel changes, it is possible to further reduce the change in business processing capacity of each department.

  <Invention of program>

  The program of the present invention is for causing a computer to function as the salesperson clustering system described above.

  The above-mentioned program or a part thereof is, for example, a magneto-optical disk (MO), a compact disk (CD), a digital versatile disk (DVD), a flexible disk (FD), a magnetic tape, or a read-only memory (ROM). , Electrically erasable and rewritable read-only memory (EEPROM), flash memory, random access memory (RAM), hard disk, solid state drive (SSD), flash disk, etc. And can be transmitted using a transmission medium such as a wired network such as LAN, MAN, WAN, Internet, Intranet, Extranet, or a wireless communication network, or a combination thereof. And It is also possible to transport and put in a carrier wave. Furthermore, the above program may be a part of another program, or may be recorded on a recording medium together with a separate program.

  As described above, according to the present invention, as a cluster variable, a feature vector having each revenue amount of a plurality of quantiles in a revenue distribution for each sales employee as a constituent element is created for each sales employee. Since cluster analysis is performed using feature vectors as input, it is possible to eliminate arbitraryness when creating cluster variables and to create cluster variables that effectively capture the characteristics of salespeople. is there.

1 is an overall configuration diagram of a salesperson clustering system according to an embodiment of the present invention. The figure which shows the structural example of the attribute table of the database which comprises the salesclerk data storage means of the said embodiment. The figure which shows the structural example of the revenue amount and the contact frequency table of the database which comprises the salesclerk data storage means of the said embodiment. The figure which shows the structural example of the feature vector table which concerns on the revenue amount of the database which comprises the salesclerk data storage means of the said embodiment. The figure which shows the profit distribution of one sales employee of the said embodiment. The figure which shows the dendrogram created by the cluster analysis in the said embodiment. The block diagram of the classification result memory | storage means of the said embodiment. The figure of the flowchart which shows the flow of the personnel change assistance process of the said embodiment. The figure which shows the structural example (state before arrangement | positioning start) of the transfer object person table of the database which comprises the data storage means for personnel changes of the said embodiment. The figure which shows the structural example (state immediately after formation of an empty frame) of the empty frame table of the database which comprises the data storage means for personnel changes of the said embodiment. The figure which shows the structural example (state immediately after formation of an empty frame) of the department table of the database which comprises the data storage means for personnel changes of the said embodiment. The figure which shows the frequency | count distribution of contact of one sales employee of the said embodiment. The figure which shows the structural example of the feature vector table which concerns on the frequency | count of contact of the database which comprises the salesclerk data storage means of the said embodiment.

  An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows the overall configuration of a salesperson clustering system 10 of the present embodiment. 2 to 4 show configuration examples of a database attribute table 30A, a revenue amount / contact count table 30B, and a feature vector table 30C relating to the revenue amount, which constitute the salesperson data storage means 30. FIG. 5 shows a profit histogram showing the profit distribution of one sales employee, FIG. 6 shows a dendrogram created by cluster analysis, and FIG. 7 shows a classification result storage means 31. The configuration of is shown. Further, FIG. 8 is a flowchart showing the flow of the personnel change support process, and FIGS. 9 to 11 show the transfer target person table 32A, empty frame table 32B of the database constituting the personnel change data storage means 32, A configuration example of the department table 32C is shown. FIG. 12 shows the contact frequency distribution of one sales employee, and FIG. 13 shows a configuration example of the feature vector table 30D related to the database contact frequency constituting the sales employee data storage means 30. ing.

  <Overall configuration of salesperson clustering system 10>

  In FIG. 1, a salesperson clustering system 10 is constituted by one or a plurality of computers, and stores processing means 20 for executing various processes relating to salesperson clustering and various data necessary for processing by the processing means 20. Sales staff data storage means 30, classification result storage means 31, personnel change data storage means 32, input means 40 such as a mouse or keyboard connected to the processing means 20, and liquid crystal connected to the processing means 20, for example. Display means 50 such as a display.

  The processing unit 20 includes an input receiving unit 21, a feature vector creating unit 22, a cluster analysis unit 23, a personnel change support unit 24, and an output unit 25. Each of the means 21 to 25 constituting these processing means 20 is a central processing unit (CPU) provided inside the computer main body constituting the salesperson clustering system 10 and one for defining the operation procedure of this CPU. Alternatively, it is realized by a plurality of programs.

  The salesperson clustering system 10 may be configured by a server / client system. In this case, the storage means 30 to 32 are installed on the server side, and the input means 40 and the display means 50 are on the client side. Installed. The main functions of the processing means 20 are installed on the server side, but some functions may be installed on the client side.

  <Configuration of input receiving means 21>

  The input receiving unit 21 executes a process of receiving various inputs by a user who operates the input unit 40. When the salesperson clustering system 10 is a server / client system, the input receiving unit 21 transmits input screen display data to the client terminal via the network (thereby, the display unit 50 of the client terminal). The input screen is displayed on the screen, and the user can input data by operating the input means 40.) The process of receiving the input data by the user transmitted from the client terminal via the network is executed. To do.

  Specifically, the input receiving means 21 executes processing for receiving input of the number of customers in charge for determining the number of higher quantiles and the step size. The number of customers in charge information is information regarding the number of customers in charge of one sales employee. When the input reception means 21 receives the input of the range of the number of customers as the information on the number of customers in charge, for example, less than 100, 100 to 199, 200 to 399, 400 to 999, 1000 or more The selection input of the user from some ranges such as the above is accepted, or the selection input of the user from, for example, about 100 people, about 500 people, about 1000 people, etc. is accepted. Further, the input receiving means 21 receives the user's input about the average number of customers in charge of each sales employee and the maximum or minimum number of customers in charge of each sales employee as the information on the number of customers in charge. Also good. Further, the input reception means 21 uses the number of customers in charge of each salesperson stored in the attribute table 30A (see FIG. 2) of the salesperson data storage means 30 by the feature vector creation means 22, or the revenue amount / contact Using the data of the frequency table 30B (refer to FIG. 3), accepting a selection input for automatically calculating the number of customers in charge (automatic calculation of the average number, maximum number, or minimum number of customers in charge of each sales employee) May be.

  Further, the input receiving unit 21 receives a selection input of the method of the first classification (first distribution selection) by the cluster analyzing unit 23. That is, the input receiving means 21 includes (P1) a classification process only for the first cluster analysis based on the profit distribution, (P2) a first cluster analysis based on the profit distribution, and a second cluster analysis based on the contact frequency distribution. A selection input as to which of the combined classification processes to be executed by the cluster analysis means 23 is accepted.

  Further, the input accepting means 21 performs detailed selection at the time of salesperson assignment or exchange by the personnel change support means 24 (if the clustering by the cluster analysis means 23 is the primary distribution selection, the secondary distribution selection The method selection input is accepted. (Q1) Method of using distance between feature vectors, (Q2) Method of using number of customers in charge, (Q3) Method of using total value of revenue amount, (Q4) Second cluster based on number of contacts Of the methods using the classification result by analysis (selectable only when P1 is selected), a selection input as to which method is to be executed by the personnel change support means 24 is accepted.

  The input receiving unit 21 receives an input of the first division number based on the first cluster analysis, or receives a selection input for automatically determining the first division number. Further, the input receiving means 21 receives a selection input for accepting an input of the second division number by the second cluster analysis when the user selects P2 or Q4, or automatically determining the second division number. Accept.

  Furthermore, the input receiving means 21 inputs, for example, information on the transfer target person (sales employee identification information about the transfer target person and department identification information about the planned transfer target department) whose transfer destination is determined by a request from an officer or the like. And the received information is stored in the transfer target person table 32A (see FIG. 9) of the personnel transfer data storage means 32.

  <Configuration of Feature Vector Creation Unit 22>

  The feature vector creating unit 22 uses the revenue amount data stored in the revenue amount / contact count table 30B (see FIG. 3) of the salesperson data storage unit 30, and uses the vertical axis as the number of customers as a cluster variable. A feature vector whose components are the calculated revenue amounts for each quantile in the revenue distribution for each sales employee (see Fig. 5), with the horizontal axis as the revenue amount raised from one customer. Is created for each salesperson, and the created feature vector for each salesperson is stored in the feature vector table 30C (see FIG. 4) related to the revenue amount of the salesperson data storage means 30 in association with the salesperson identification information. Is to execute. The revenue distribution for each sales employee shown in FIG. 5 shows a bias in the revenue amount of active customers within a certain period. The details of the feature vector creation process using the revenue amount data stored in the revenue amount / contact count table 30B (see FIG. 3) will be described later (see step S2 in FIG. 8).

  At this time, in the present embodiment, the feature vector creating unit 22 exemplifies the first quartile, the second quartile, and the third quartile in the profit distribution for each salesperson (see FIG. 5). And the at least one higher quantile that subdivides the larger revenue range than the third quartile with the largest revenue amount among these three quartiles. A process of creating a four-dimensional or higher feature vector having a revenue amount as a constituent element for each salesperson is executed. Note that the plurality of quantiles used for creating the feature vector related to the revenue amount is not limited to this, and may be any quantile instead of the quartile, for example. FIG. 5 is a diagram for explaining the positional relationship of a plurality of quantiles, and the scale of the revenue amount on the horizontal axis in this revenue distribution is not accurate.

  Further, the feature vector creation means 22 uses the number of customers in charge by the user input received by the input reception means 21, or the attribute table 30A (see FIG. 2) of the salesperson data storage means 30, or the amount of revenue / contact count. The number of customers in charge is automatically calculated using the data stored in the table 30B (see FIG. 3), and the correspondence between the number of customers in charge and the number of higher quantiles and the step size is determined in advance. A process of determining the number and step size of upper quantiles provided by subdividing is executed. The correspondence relationship between the number of customers in charge, the number of higher quantiles, and the step size may be described in the program, stored in a memory (quantitative point storage means not shown), You may make it read it. Specifically, for example, when the number of customers in charge ranges from 200 to 599 or about 400, the quartile, the top 10% point, and the top 5% point If the top 1% point and the top 0.5% point are used, and the number of customers in charge is in the range of 1000 to 4999 people, or about 2000 people, in addition to these, the top 0.1% point is used. It is a correspondence relationship such as use.

  Further, the feature vector creating means 22 adds the maximum value of the revenue amount raised from each customer in charge by one sales employee in addition to the revenue amounts of the plurality of quantiles as described above, At least one additional element selected from the total value of the revenue amount raised by each sales representative from each customer and the average value of the revenue amount raised by each sales representative from each customer Processing for creating a feature vector for each salesperson may be executed. The selection to insert such additional elements into the constituent elements of the feature vector is received by the input receiving means 21.

  Further, the feature vector creating means 22 uses the data on the number of contacts stored in the revenue amount / contact count table 30B (see FIG. 3) of the salesperson data storage means 30, and uses the vertical axis as the cluster variable. And the horizontal axis represents the number of contacts to one customer, and the number of contacts at a plurality of quantiles in the distribution of contact numbers for each salesperson (see FIG. 12) is obtained. A vector is created for each salesperson, and the created feature vector for each salesperson is stored in the feature vector table 30D (see FIG. 13) related to the number of contacts in the salesperson data storage means 30 in association with the salesperson identification information. The process is executed. The details of the feature vector creation processing using the contact count data stored in the revenue amount / contact count table 30B (see FIG. 3) are the same as the feature vector creation processing related to the revenue amount (described later). (See step S2 in FIG. 8). FIG. 12 is a diagram for explaining the positional relationship between a plurality of quantiles, and the scale of the number of contacts on the horizontal axis in this contact number distribution is not accurate.

  When creating the feature vector related to the number of contacts, the feature vector creating unit 22 uses a plurality of quantile points, for example, quartile and Although at least one higher quantile is used, it is not always necessary to use the same quantile as in the process of creating the feature vector related to the revenue amount, and a different quantile may be used. Note that the number of customers in charge information is the same as in the process of creating a feature vector related to the amount of revenue, so when creating a feature vector related to the number of contacts, A process of determining the number and step widths of the upper quantiles provided in accordance with the correspondence relationship between the number of step points and the step width is executed.

  <Configuration of Cluster Analysis Unit 23>

  The cluster analysis unit 23 uses the feature vector for each salesperson created by the feature vector creation unit 22 as input, classifies a plurality of salespeople by cluster analysis, and stores the classification result in the classification result storage unit 31 (see FIG. 7). The process is executed. At this time, as shown in FIG. 7, the cluster analysis means 23 stores the classification result by associating the cluster number or cluster feature value assigned to each cluster with the salesperson identification information about the salesperson classified into each cluster. The data is stored in the means 31.

  The cluster number is a number assigned in accordance with the degree of excellence in sales performance of each cluster (usually a serial number). For example, when the number of divisions is 7 (see FIG. 6), the business results In accordance with the degree of excellence, for example, 1, 2, 3, 4, 5, 6, 7 and so on are assigned in order from the cluster with the lowest grade. When assigning a cluster number to each cluster, which cluster is the cluster of salespeople with excellent sales results, for example, the top of multiple salespeople (all salespeople) classified into the same cluster It can be determined by the average value of the revenue amount of 0.5 points, etc., and it is preferable to automatically assign the cluster number based on such a predetermined judgment index, but it is not necessarily limited to automatic numbering Instead, the user may manually assign a cluster number to each cluster. In this case, the input receiving unit 21 receives the numbering input. The cluster feature amount is a quantity representing the degree of excellence in sales performance of each cluster. For example, the top 0.5 revenue amount of a plurality of sales employees (all sales employees) classified into the same cluster. In general, it is a non-linear numerical sequence.

  The cluster analysis of the present invention may be based on an existing method. For example, clustering is performed using distance calculation such as Euclidean distance using the feature vector of each salesperson created by the feature vector creation means 22 as a cluster variable, thereby forming the shape of the customer portfolio (ie, as shown in FIG. 5). Sales representatives with similar revenue distribution shapes are grouped (see FIG. 6). Therefore, the cluster analysis means 23 can be configured using a library prepared in an existing language such as R language or Python. Further, instead of using an existing library, coding may be performed from one, but even in that case, the clustering algorithm may be an existing algorithm.

  More specifically, the cluster analysis unit 23 inputs a feature vector for each salesperson having the revenue amount created by the feature vector creation unit 22 as a component and inputs a plurality of salespeople to the first by the first cluster analysis. A first classification process is performed in which the classification results are classified into clusters and the classification results are stored in the classification result storage unit 31 (see FIG. 7). For example, as shown in FIGS. 6 and 7, when the first division number = 7, the cluster numbers 1, 2, 3, 4, 5, 6, and 7 are automatically assigned in order from the lowest sales results. Shake. Note that the user may manually assign the cluster number.

  Further, the cluster analysis means 23 receives a feature vector for each salesperson having the number of contacts created by the feature vector creation means 22 as a component and inputs a plurality of salespersons into a second division number cluster by the second cluster analysis. The second classification process is executed in which the classification result is stored in the classification result storage means 31 (see FIG. 7). The second division number by the second cluster analysis may be the same as or different from the first division number by the first cluster analysis. For example, as shown in FIG. 7, when the second number of divisions = 3, in order from the lowest sales results (which numerical value leads to good business results is the same as in the first cluster analysis. Similarly, it is determined in advance and automatically determined based on the size of the numerical value.) Cluster numbers 1, 2, and 3 are automatically assigned. Note that the user may manually assign the cluster number. However, this second classification process is executed when the user selection of P2 and Q4 described above (see step S1 in FIG. 8) is received by the input receiving means 21.

  Further, when the cluster analysis unit 23 receives the above-described user selection of P2 (see step S1 in FIG. 8) by the input reception unit 21, the first classification result obtained by the above-described first classification process. And the second classification result obtained by the second classification process described above are used to classify a plurality of sales employees into clusters having a number obtained by multiplying the first division number and the second division number. Then, a composite classification process for storing the classification results in the classification result storage means 31 (see FIG. 7) is executed. For example, as shown in FIG. 7, when the first division number = 7 and the second division number = 3, 7 × 3 = 21 clusters are formed. No. 21. This composite classification process need not be executed when the above-described user selection of Q4 (see step S1 in FIG. 8) is received by the input receiving means 21.

  <Configuration of personnel change support means 24>

  The personnel change support means 24 supports the relocation work between sales departments of the sales staff, and the data storage means 32 for personnel changes (see FIGS. 9 to 11) shows the result of the sales staff assignment and exchange accompanying the personnel change. The personnel change support process to be memorized is executed.

  More specifically, the personnel change support means 24 associates with the sales person identification information about the transfer-side sales staff placed in each department when the sales staff is placed in each department. The cluster number or cluster feature value stored in FIG. 7) is associated with the salesperson identification information about the salesperson on the transfer side who has formed an empty frame in each department due to the transfer from each department to another department. A person selection process for matching the cluster number or the cluster feature amount stored in the classification result storage unit 31 is executed. Further, the personnel change support means 24 is a cluster stored in the classification result storage means 31 in association with salesperson identification information about two salespeople to be exchanged when exchanging salespeople between two departments. A person selection process for matching numbers or cluster features is executed.

  Further, the personnel change support means 24, when arranging the sales staff in each department, the cluster number or cluster feature quantity of the transfer-side sales staff, the cluster number or cluster feature quantity of the plurality of outgoing side matches. When there are sales employees, or when there are multiple transfer-side sales employees whose cluster numbers or cluster feature quantities match the cluster number or cluster feature quantity of the transfer-side sales staff, or between two departments When there are a plurality of replacement salespersons whose cluster numbers or cluster feature quantities match when exchanging salespersons in, the methods Q1, Q2, Q3 which received the user's selection input by the input reception means 21 Q4 (see step S1 in FIG. 8 to be described later), the detailed selection of the salesperson (clustering by the cluster analysis means 23) If the first primary sorting selected and secondary distribution selected.) To execute the processing of.

  The method Q1 is a method of using the distance between feature vectors, and the personnel change support means 24 has a feature vector table 30C (FIG. 4) of the salesperson data storage means 30 when there are a plurality of candidates (salespeople). The process of selecting the salesperson with the shortest distance between the two calculated using the feature vector related to the revenue amount stored in (see) is executed.

  The method Q2 is a method of using the number of customers in charge, and the personnel change support unit 24 stores the attribute table 30A (see FIG. 2) of the salesperson data storage unit 30 when there are a plurality of candidates (salespeople). A process of selecting the salesperson with the closest number of stored customers in charge is executed.

  The method Q3 is a method of using the total value of the revenue amount, and the personnel change support means 24 has an attribute table 30A (see FIG. 2) of the salesperson data storage means 30 when there are a plurality of candidates (salespeople). ) Is executed to select a salesperson with the closest total value of the revenue amount stored in ().

  In the methods Q2 and Q3, the number of customers in charge and the total value of revenue are used, but other data stored in the attribute table 30A (see FIG. 2) of the salesperson data storage means 30, for example, , Age, years of service, etc. may be used to execute a process of selecting a salesperson with the closest value.

  The method Q4 is a method of using the classification result obtained by the second cluster analysis based on the number of contacts, and the personnel change support unit 24 uses the classification result storage unit 31 (FIG. 5) when there are a plurality of candidates (salespeople). 7), a process of selecting a salesperson whose cluster numbers or cluster feature quantities of the second classification process classified by the second cluster analysis based on the number of contacts stored in the second match analysis or close is selected. This method Q4 can be selected only when the user's selection input of P1 (see step S1 in FIG. 8 described later) is received by the input receiving means 21, and the user selects the P2 composite classification process. In this case, since the classification result by the second cluster analysis has already been utilized in the composite classification process, this method Q4 cannot be selected.

  <Configuration of output means 25>

  The output means 25 is the result of classification by the cluster analysis means 23 (contents of data stored in the classification result storage means 31 (see FIG. 7)), and the result of salesperson assignment or exchange by the personnel change support means 24 (personnel affairs). Transfer data storage means 32 (contents of data stored in FIGS. 9 to 11) is displayed on a screen, printed, transmitted to another system (not shown) via a communication line, and recorded on a recording medium. The process of storing or outputting the sound is executed. When the salesperson clustering system 10 is a server / client system, the output means 25 executes a process of transmitting display data of the result output screen to the client terminal via the network. As a result, the result output screen is displayed on the screen of the display means 50 of the client terminal.

  <Configuration of each storage means 30-32>

  The salesperson data storage means 30, the classification result storage means 31, and the personnel change data storage means 32 are each constituted by a database in this embodiment, but the data holding / management form is not a database but a file. It is good also as a system. For example, when the salesperson clustering system 10 is a server / client system, or when it is premised on simultaneous use of a plurality of users, a database server may be installed (in this case, the processing means 20 In the case where the salesperson clustering system 10 is a stand-alone system configured by a single computer, a data file of spreadsheet software may be used. . The constituent devices of the storage means 30 to 32 are preferably realized by, for example, a hard disk, a solid state drive (SSD), or the like. However, as long as there is no problem in storage capacity, access speed, etc., DVD, flash, Other recording media such as a memory may be used.

  The salesperson data storage means 30 includes an attribute table 30A (see FIG. 2), a revenue amount / contact count table 30B (see FIG. 3), a feature vector table 30C (see FIG. 4) relating to the revenue amount, and a contact count. This feature vector table 30D (see FIG. 13) is provided.

  As shown in FIG. 2, the attribute table 30A includes salesperson identification information, age, years of service, department identification information of departments, titles, number of customers in charge, and information from each customer. The maximum value among the revenue amounts, the total value of the revenue amounts raised from each customer, the average value of the revenue amounts raised from each customer, and the like are stored in association with each other. Although not shown in FIG. 2, the number of years that have elapsed since the current department was placed may be stored.

  As shown in FIG. 3, the revenue amount / contact count table 30B stores salesperson identification information, customer identification information, revenue amount, and contact count in association with each other. Note that the revenue amount and the contact count may be separate tables, in which case the revenue amount and the contact count are stored in association with the salesperson identification information and the customer identification information. Further, although illustration in FIG. 3 is omitted, when a salesperson performs a business visit to a customer, the visit time may be stored in association with the salesperson identification information and the customer identification information. The revenue amount and contact count stored in the revenue amount / contact count table 30B are data relating to active customers (customers who have made or sold at least once within a certain period) within a certain period (for example, one year). The same applies to the number of customers in charge and the maximum value, total value, average value, etc. of the revenue amount stored in the attribute table 30A (see FIG. 2).

  As shown in FIG. 4, the feature vector table 30C related to the revenue amount includes a plurality of quantiles (in the example of FIG. 4, the first quartile, The second quartile, the third quartile, the top 10% point, the top 5% point, the top 1% point, and the top 0.5% point) are stored in association with the salesperson identification information. To do. In the example of FIG. 4, the feature vector is a 7-dimensional vector.

  As shown in FIG. 13, the feature vector table 30D related to the number of contacts has a plurality of quantile points (in the example of FIG. 13, the first quartile, The second quartile, the third quartile, the top 10% point, the top 5% point, the top 1% point, and the top 0.5% point) are stored in association with the salesperson identification information. To do. In the example of FIG. 13, the feature vector is a 7-dimensional vector.

  As shown in FIG. 7, the classification result storage unit 31 uses, as data indicating the classification result by the cluster analysis, the cluster number of the first classification process classified by the first cluster analysis based on the salesperson identification information and the profit distribution. Alternatively, cluster feature quantities (cluster numbers 1 to 7 in the example of FIG. 7), cluster numbers or cluster feature quantities of the second classification process classified by the second cluster analysis based on the contact frequency distribution (FIG. 7). In the example of FIG. 7, the cluster number or the cluster feature amount of the combined classification process combining the classification results obtained by the first and second cluster analysis (numbers 1 to 21 in the example of FIG. 7). Are stored in association with each other.

  The personnel change data storage means 32 includes a transfer subject table 32A (see FIG. 9), an empty frame table 32B (see FIG. 10), and a department table 32C (see FIG. 11).

  As shown in FIG. 9, the transfer target person table 32 </ b> A includes, as data of a transfer target person (transfer-side sales staff placed in any department), sales employee identification information about the transfer target person, The cluster number or cluster feature amount (in the example of FIG. 9, the first to seventh cluster numbers), the department identification information about the scheduled transfer destination department, and the department identification information about the confirmed transfer destination department are stored in association with each other. Is. Among these, the department identification information regarding the department to which the planned change is made is input information of the user received by the input receiving means 21, for example, when a specific salesperson is relocated to a specific department upon request from an officer. On the other hand, this department identification information is “undecided” because, for example, a salesperson who satisfies the condition that the continuous enrollment period in the same department reaches the maximum number of years is stored in the salesperson data storage means 30. This is the case where the data is automatically extracted using the data of the attribute table 30A (see FIG. 2).

  As shown in FIG. 10, the empty frame table 32B includes, as the empty frame data formed in each department, the department identification information about the department in which the empty frame is formed, and the salespersons in the empty frame (out of the department). The cluster number or cluster feature quantity (in the example of FIG. 10, the cluster numbers 1 to 7 in the example of FIG. 10), the salesperson identification information about the salesperson who was in the vacant frame, and the vacant frame. The salesperson identification information about the salesperson (the salesperson on the transfer side arranged in the department) is stored in association with each other.

  As shown in FIG. 11, the department table 32 </ b> C includes department identification information, a permissible range of increase / decrease in the number of departments associated with personnel changes, and the entire department as a result of the formation of an empty frame. Decrease of the sum of cluster numbers or cluster feature values at the same time, increase of the sum of cluster numbers or cluster feature values for the entire department due to the sales staff assignment, and the cluster number of the entire department before and after personnel changes The increase or decrease of the sum or the sum of the cluster feature values is stored in association with each other.

  <Flow of cluster analysis and personnel change support>

  In this embodiment, the salesperson clustering system 10 performs the salesperson classification process by cluster analysis and the personnel change support process using the classification result as follows.

  In FIG. 8, first, when a user inputs information necessary for cluster analysis and personnel change support processing, this input information is received by the input receiving means 21 (step S1).

  Specifically, the input accepting unit 21 accepts input of the number of customers in charge for determining the number of higher quantiles and the step size (step S1). For example, the number of customers in the number of customers in charge is less than 100, 100 to 199, 200 to 399, 400 to 999, 1000 or more, etc. Select and input the range. The input accepting unit 21 may accept a user selection for automatically calculating the number of customers in charge. In this case, as will be described later, the feature vector creation unit 22 automatically calculates the number of customers in charge. Is executed.

  Further, the input receiving unit 21 receives a selection input of the method of the first classification (first distribution selection) by the cluster analyzing unit 23 (step S1). That is, the input receiving means 21 performs (P1) a classification process only for the first cluster analysis based on the profit distribution, (P2) a first cluster analysis based on the profit distribution, and a second cluster analysis based on the contact count. A selection input as to which of the combined composite classification processes is to be executed by the cluster analysis means 23 is accepted.

  Further, the input accepting means 21 makes a detailed selection at the time of salesperson assignment or exchange by the personnel change support means 24 (if the clustering by the cluster analysis means 23 is the primary distribution selection, the secondary distribution selection Is received) (step S1). (Q1) Method of using distance between feature vectors, (Q2) Method of using number of customers in charge, (Q3) Method of using total value of revenue amount, (Q4) Second cluster based on number of contacts Of the methods using the classification result by analysis (selectable only when P1 is selected), a selection input as to which method is to be executed by the personnel change support means 24 is accepted.

  Further, the input receiving unit 21 receives an input of the first division number by the first cluster analysis based on the profit distribution, or receives a selection input for automatically determining the first division number by the cluster analysis unit 23. (Step S1). Further, the input receiving unit 21 receives an input of the second division number based on the second cluster analysis based on the contact frequency distribution when the user selects P2 and Q4, or the second division number is determined as the cluster analysis unit. 23, a selection input for automatic determination is accepted (step S1). For example, the user inputs the first division number = 7, the second division number = 3, and the like.

  Furthermore, the input receiving means 21 inputs, for example, information on the transfer target (sales employee identification information about the transfer target person and department identification information about the planned transfer target department) whose transfer destination is determined by an officer's request or the like. The received information is stored in the transfer target person table 32A (see FIG. 9) of the personnel transfer data storage means 32 (step S1). As a result, a record of the transfer target person whose transfer destination is determined is created in the transfer target person table 32A, and the “department identification information of the planned transfer target department” column of the transfer target person table 32A is shown in FIG. As shown, for example, the department α, β, γ, etc., to which the scheduled change is made are stored. At this stage, since the process by the cluster analysis means 23 is not performed, the cluster number or the like is not stored in the “cluster number or cluster feature amount of transfer target person” column, and personnel Since the processing by the transfer support means 24 is not performed, the section identification information is not stored in the “department identification information of the determined transfer destination department” column.

  Subsequently, a feature vector for each salesperson is created by the feature vector creating means 22 (step S2). At this time, if the user selection for automatically calculating the number of customers in charge is accepted by the input accepting means 21 in step S1 described above, the attribute table of the salesperson data storage means 30 is obtained by the feature vector creating means 22. The number of customers in charge of each salesperson stored in 30A (see FIG. 2) is used, or the number of customers in charge information (each salesperson's information) is obtained using the data in the revenue / contact count table 30B (see FIG. 3). An automatic calculation process of an average number, a maximum number, or a minimum number of customers in charge is executed.

  The feature vector creating unit 22 uses the number of customers in charge by the user input received by the input receiving unit 21 in the above-described step S1 or uses the number of customers in charge calculated automatically to determine a charge in advance. In accordance with the correspondence relationship between the number of customers information, the number of higher quantiles, and the step size, the number and step widths of the upper quantile points to be provided are determined (step S2). For example, if the number of customers in charge is in the range of 200 to 599 or about 400, three quartiles, the top 10%, and the top 5% are used as a plurality of quantiles to be used. Points, top 1% point, top 0.5% point are determined.

  Then, using the profit amount data stored in the revenue amount / contact count table 30B (see FIG. 3) of the salesperson data storage means 30 by the feature vector creation means 22, the vertical axis indicates the number of customers as a cluster variable. The horizontal axis represents the amount of revenue raised from one customer, and each revenue amount for multiple quantiles in the revenue distribution for each salesperson (see Fig. 5) is calculated, and each calculated revenue amount is used as a component. A feature vector is created for each salesperson (step S2).

  More specifically, the feature vector creation unit 22 sorts the records of the revenue amount / contact count table 30B in FIG. 3 in descending order of the revenue amount. Thereby, what corresponds to the profit histogram (refer FIG. 5) which shows the profit distribution for every sales employee was formed. Here, for example, if the number of customers in charge of salesperson identification information = B0001 is 400, the revenue amount of the first quartile (lower 25% point) is 400 people × 25% = 100. Since it is a person, the revenue amount raised from the customer (customer identification information = C0007676) of the 100th record from the bottom of the revenue amount is 1 million yen. The record of the 100th person from the bottom of the revenue amount is not the record of the 300th person from the top of the revenue amount, but the record of the 301st person, and the record of the 101st person from the bottom is the record of the 300th person from the top. The revenue amount raised from the customer of the 101st record from the bottom (customer identification information = C00456732) = 1 million yen may be adopted, or the revenue amount of the 100th customer from the bottom and the revenue of the 101st customer An average with the amount of revenue may be taken. Further, for example, when a fraction is obtained such as 355 people × 25% = 88.75 people, a weighted average of the revenue amounts on both sides may be taken.

  Similarly, since the revenue amount (median) of the second quartile is 400 people × 50% = 200 people, it is raised from the customer (customer identification information = C00342136) of the 200th record from the bottom of the revenue amount. Revenue amount = 23,000 yen. Since the revenue amount of the third quartile is 400 people x 75% = 300 people, the revenue amount raised from the customer of the 300th record from the bottom of the revenue amount (customer identification information = C00050055) = 350,000 yen It becomes.

  In addition, since the top 0.5% revenue amount is 400 × 0.5% = 2, the revenue amount raised from the customer of the second record from the top of the revenue amount (customer identification information = C0361346) = 7.9 million yen. Similarly, since the revenue amount of the top 1% point is 400 people × 1% = 4 people, the revenue amount raised from the customer (customer identification information = C03001534) of the fourth record from the top of the revenue amount = 673 thousand Yen. Since the top 5% revenue amount is 400 people x 5% = 20 people, the revenue amount raised from the 20th record customer (customer identification information = C00349749) from the top of the revenue amount is 33.33 million yen. . Since the top 10% revenue amount is 400 people x 10% = 40 people, the revenue amount raised from the 40th record customer (customer identification information = C00000139) from the top of the revenue amount is 1.15 million yen. .

  Then, the feature vector creating means 22 creates a feature vector having each revenue amount of the plurality of quantiles thus obtained as a constituent element, and associates the created feature vector with the salesperson identification information. The data is stored in the feature vector table 30C (see FIG. 4) relating to the revenue amount of the employee data storage means 30 (step S2).

  When the user selection input of P2 and Q4 is received by the input receiving means 21 in step S1 described above, the revenue vector / contact count table 30B (see FIG. 3), using the contact count data stored in FIG. 3 as a cluster variable, the vertical axis represents the number of customers, and the horizontal axis represents the number of contacts for one customer (see FIG. 12). The number of contacts for each of the plurality of quantiles is determined, and a feature vector having the determined number of contacts as a constituent element is created for each salesperson (step S2).

  More specifically, the records of the revenue amount / contact count table 30B in FIG. 3 are rearranged not in the revenue amount but in order of increasing or decreasing the number of contacts. As a result, a histogram corresponding to the contact frequency distribution for each salesperson (see FIG. 12) is formed. In this state, the number of times of each contact of a plurality of quantiles is obtained in the same manner as in the case of the revenue amount. Then, a feature vector having each contact count of the obtained plurality of quantiles as a constituent element is created, and the created feature vector is associated with the salesperson identification information, and the feature related to the contact count of the salesperson data storage means 30 It is stored in the vector table 30D (see FIG. 13) (step S2).

  Subsequently, the salesperson created by the feature vector creation means 22 in the above-described step S2 by the cluster analysis means 23 and stored in the feature vector table 30C (see FIG. 4) relating to the revenue amount of the salesperson data storage means 30. Each feature vector is input and a plurality of sales employees (here, all sales employees belonging to a group such as a company having a plurality of departments) are classified into the first division number by the first cluster analysis. The classification result is stored in the classification result storage means 31 (see FIG. 7) (step S3). For example, as shown in FIG. 6, when the first division number = 7, the first to seventh cluster numbers are automatically assigned to the created seven clusters in order from the lowest sales results. Then, the cluster number for the cluster into which each salesperson is classified is associated with the salesperson identification information for each salesperson and stored in the classification result storage means 31 of FIG.

  Further, when the user selection input of P2 and Q4 is received by the input receiving means 21 in the above-described step S1, it is created by the cluster vector analyzing means 23 by the feature vector creating means 22 in the above-described step S2, and the salesperson data. A plurality of salespersons (here, having a plurality of departments) by the second cluster analysis with the feature vector for each salesperson stored in the feature vector table 30D (see FIG. 13) related to the number of contacts in the storage means 30 as an input. All sales employees belonging to a group such as a company are classified into the second division number, and the classification result is stored in the classification result storage means 31 (see FIG. 7) (step S3). For example, as shown in FIG. 7, when the second number of divisions = 3, the first to third cluster numbers are automatically assigned to the three created clusters in order from the lowest sales results. Then, the cluster number for the cluster into which each salesperson is classified is associated with the salesperson identification information for each salesperson and stored in the classification result storage means 31 of FIG.

  Furthermore, when the user selection input of P2 is received by the input receiving unit 21 in the above-described step S1, the first classification obtained by the first classification process based on the above-described revenue amount is performed by the cluster analyzing unit 23. By combining the result and the second classification result obtained in the second classification process based on the number of contacts described above, a plurality of sales employees (here, all belonging to an organization such as a company having a plurality of departments) Are classified into a number of clusters obtained by multiplying the first division number and the second division number, and the classification result is stored in the classification result storage means 31 (see FIG. 7) (step S3). ). For example, as shown in FIG. 7, when the first number of divisions = 7 and the second number of divisions = 3, 7 × 3 = 21 clusters are formed, so the created 21 clusters On the other hand, the first to 21st cluster numbers are automatically assigned in order from the lowest sales results, and the cluster numbers for the clusters into which each salesperson is classified are associated with the salesperson identification information for each salesperson. The result is stored in the classification result storage means 31 in FIG.

  After that, the personnel change support means 24 starts a process (personnel change support process) for supporting the relocation work between the departments of the sales staff. First, the person to be transferred (sales staff) is extracted from each department and extracted. In addition to executing the process of storing the information of the transfer target person, at the same time, forming a vacant frame for accepting the transfer target person in each department and executing the process of storing the information of the formed vacant frame (step) S4).

  In the following, the data to be stored in the revenue amount / contact count table 30B (see FIG. 3) of the salesperson data storage means 30 for the relocation of a salesperson who has already belonged to an organization (such as a company) for a certain period of time. New employees (persons who become sales employees) who do not have a cluster analysis cannot be performed, so they are not considered. New employees are assigned to each department by a method different from the present invention.

  Retired employees (those who were sales employees) are considered to create vacancies in each department. However, they have retired or have already retired, so they are not exchanged. Therefore, if the increase in the number of new employees is not taken into account as described above, the total number of employees (for the entire organization) will be reduced by the number of retired employees before and after personnel changes.

  Further, in the department table 32C (see FIG. 11) of the personnel change data storage means 32, information on each department α, β, γ,... Before and after, for example, a department that maintains the number of people, a department that allows the increase or decrease of the number of people in the range of -2 to +1, and the like are registered in advance. This registration is received by the input receiving means 21.

  More specifically, by using the data stored in the attribute table 30A (see FIG. 2) of the salesperson data storage means 30 by the personnel change support means 24, for example, the continuous enrollment period of the current department is predetermined. Those who satisfy the conditions such as reaching the maturity are extracted from each department α, β, γ,. This extraction process is executed for all departments, that is, for the entire company (for the entire organization). Note that all or part of this extraction processing may be performed manually by the user, and the input of the extraction result in that case is received by the input receiving means 21.

  With respect to the extracted transfer target person, the personnel transfer support means 24 creates a new record in the transfer target person table 32A (see FIG. 9) of the personnel change data storage means 32, and the transfer record shows the transfer target person's name. Stores salesperson identification information. Since the transfer target person extracted here has not been determined, “undecided” is stored in the “department identification information of the planned transfer target department” column of the transfer target person table 32A. On the other hand, with respect to the transfer target person whose transfer destination has been determined, the record has already been created by the input receiving means 21 in step S1 described above, and the sales employee identification information received by the input receiving means 21 and the department of the planned transfer destination Since the department identification information is stored, another record is created in the transfer target person table 32A by the personnel transfer support means 24.

  Then, the personnel change support means 24 acquires the cluster number or cluster feature amount of the transfer target person from the classification result storage means 31 (see FIG. 7) using the salesperson identification information as a key, and the transfer target person table 32A (FIG. 9). In the “cluster number or cluster feature quantity of transfer target” column. At this stage, the “department identification information of the confirmed transfer destination department” column is blank or NULL. Note that the cluster number or cluster feature amount acquired from the classification result storage unit 31 (see FIG. 7) is obtained based on the profit distribution when the input selection unit 21 receives the user selection input of P1 in step S1 described above. When the cluster selection number or the cluster feature amount (any one of No. 1 to No. 7 in the example of FIG. 7) of the first classification process classified by the cluster analysis of 1 is received and the user selection input of P2 is accepted , A cluster number or cluster feature value (in the example of FIG. 7, any one of Nos. 1 to 21) of the combined classification process that combines the classification results of the first and second cluster analysis.

  In addition, since the transfer target person whose transfer destination is determined, the transfer target person extracted from each department in the success / failure determination of the condition, and the retired person will form an empty frame in each department, personnel transfer support means 24 Thus, a new record is created in the empty frame table 32B (see FIG. 10) of the personnel change data storage means 32, and the department identification information about the department in which the empty frame is formed and the empty frame are included in the generated record. Stores sales employee identification information for sales employees (transferees whose transfer destinations have been determined, undecided transfer targets, and retired employees).

  Then, the personnel change support means 24 uses the salesperson identification information to identify the cluster number or cluster feature amount of the salesperson (the transferee whose transfer destination has been decided, the tentative transferee, and the retired employee) in the vacant space. Is obtained from the classification result storage means 31 (see FIG. 7) as a key and stored in the “cluster number or cluster feature amount of the salesperson in the empty frame” column of the empty frame table 32B (see FIG. 10). At this stage, since no transfer target person is arranged in the vacant frame, the “salesperson identification information of the arranged salesperson” column is blank or NULL. The cluster number or cluster feature amount acquired from the classification result storage unit 31 (see FIG. 7) is the same as that in the transfer target person table 32A (see FIG. 9).

  Subsequently, the personnel transfer support means 24 places the transfer target person whose transfer destination has been determined in the vacant frame of the transfer destination department (step S5). Therefore, the transfer target person whose transfer destination is undecided (any department is acceptable) is not yet assigned.

  Here, for simplification of explanation, the following symbols are assigned to the transfer target person and the empty frame. If the person to be transferred is T, then the cluster number of the person to be transferred is placed, and the parenthesis of the planned transfer destination or transfer destination is described in parentheses after that, for example, become that way.

T7 (δ): Changed subject whose cluster number is determined to be transferred to the department δ 7 T5 (ε): Changed subject whose cluster number is determined to be transferred to the department ε T6 ( TBD (undecided): Transfer destination whose cluster number is 2 and whose destination is undecided (any department is acceptable).

  In addition, for the vacant slots formed in each department, if the cluster number of the salesperson who was in the vacant slot (the salesperson who formed the vacant slot by transfer) is placed next to the department, for example: Become.

α7: Empty frame with cluster number 7 formed in department α β3: Empty frame with cluster number 3 in department β γ2: Empty frame with cluster number 2 formed in department γ

  Therefore, in the transfer target person table 32A of the personnel change data storage means 32, in the example of FIG. 9, T6 (α), T7 (γ), T1 (β), T2 (undecided), T1 (undetermined), T5 In the example shown in FIG. 10, α7, α3, α5, α2, β1, β2, β1, γ4, γ6, and γ3 are stored in the empty frame table 32B. In the following, a description that does not assume the state illustrated in FIGS. 9 and 10 will be given.

  The personnel change support means 24 first arranges the transfer target person whose transfer destination is determined to the department α in the department α, and then allocates the transfer target person whose transfer destination is determined to the department β to the department β. In other words, the transfer destination is arranged in the order of the departments α, β, γ,.

  The processing contents of the case where the vacant frame of the department α is sufficient for the transfer target person whose transfer destination is determined to the department α are as follows.

  (Case 1) For example, T7 (α), T6 (α), and T4 (α) are transfer target persons who are determined to be transferred to the department α, and the empty frames of the department α are α7, α7, α5. , Α3 and α2.

  At this time, the personnel transfer support means 24 places the transfer target person T7 (α) in the empty frame α7 whose cluster number is 7 and matches. That is, the cluster number or cluster feature amount of the transfer target person T7 (α) who is the transfer side salesperson is matched with the cluster number or cluster feature amount of the salesperson who was in the vacant space α7 which is the transfer side salesperson. Perform placement processing.

  However, there are two empty frames α7. Therefore, the personnel change support means 24 performs detailed selection. If the first selection using the cluster number or cluster feature amount, which is the classification result by cluster analysis, is the primary distribution selection, this detailed selection is the secondary distribution selection. The two empty frames α7 both have a cluster number of 7, and may be arranged in either α7. However, the two empty frames α7 are randomly arranged in one α7 and the other α7 is empty. When left in the frame, in the subsequent processing, a plurality of candidates may be generated as transfer target persons to be placed on the remaining empty frame α7. In that case, the remaining empty frame α7 Since the closest person to be transferred is selected and placed, in order to optimize the selection at that time, detailed selection is performed while avoiding random placement.

  Specifically, when the user selects Q1 in step S1 described above (see FIG. 8), the personnel change support unit 24 uses the salesperson identification information as a key to set the revenue amount in the salesperson data storage unit 30. From the feature vector table 30C (see FIG. 4), the feature vector relating to the revenue amount of the transfer target person T7 (α) and the two vacant frames α7 (sales employees who were in the vacant frame α7) is acquired, and the acquired feature vectors are The distance between the vectors of T7 (α) and the respective empty frames α7 is calculated, the empty frame α7 having the shorter distance (the closer one) is selected, and the empty frame of the personnel change data storage means 32 is selected. In the table 32B (see FIG. 10), the salesperson identification information of the salesperson who is the transfer target person T7 (α) arranged there is stored in the record of the selected empty frame α7.

  Further, when the user selects Q2 in step S1 described above (see FIG. 8), the personnel transfer support means 24 uses the salesperson identification information as a key in the attribute table 30A (FIG. 2) of the salesperson data storage means 30. The number of customers in charge of the transfer target person T7 (α) and the two vacant frames α7 (sales staff who were in the vacant frame α7) is obtained from the transfer target person T7 (α) and the number of customers in charge of T7 (α) The number of customers in charge is compared, and the number of customers in charge is selected as the vacant frame α7 closer to T7 (α), and the selected one is selected in the vacant frame table 32B (see FIG. 10) of the personnel change data storage means 32 The salesperson identification information of the salesperson who is the transfer target person T7 (α) arranged there is stored in the record of the vacant frame α7.

  Further, when the user selects Q3 in step S1 described above (see FIG. 8), the personnel change support means 24 uses the salesperson identification information as a key to set the attribute table 30A (FIG. 2) of the salesperson data storage means 30. Reference) to obtain the total value of the revenue amount of the transfer target person T7 (α) and the two vacant frames α7 (sales staff who were in the vacant frame α7), and the total value of the revenue amount of T7 (α) The total value of the revenue amount of the vacant frame α7 is compared, the vacant frame α7 whose total value of the revenue amount is closer to T7 (α) is selected, and the vacant frame table 32B of the personnel change data storage means 32 (FIG. 10). In the reference), the salesperson identification information of the salesperson who is the transfer target person T7 (α) arranged there is stored in the record of the selected empty frame α7.

  Further, when the user selects Q4 in step S1 described above (see FIG. 8: In this case, the user selects P1 and does not select P2), personnel change support means. 24 obtains the cluster number or cluster feature quantity of the second classification process classified by the second cluster analysis based on the number of contacts from the classification result storage means 31 (see FIG. 7) using the salesperson identification information as a key. , The cluster number or cluster feature amount of the second classification process of T7 (α) is compared with the cluster number or cluster feature amount of the second classification process of each empty frame α7, and the cluster of the second classification process The empty frame α7 whose number or cluster feature value matches or is closer is selected and selected in the empty frame table 32B (see FIG. 10) of the personnel change data storage means 32. Record remnant α7 towards the stores sales employee identification salesman is Moved subject T7 disposed therein (alpha).

  Then, even when the user selects any of Q1 to Q4 in step S1 described above (see FIG. 8), the personnel change support means 24 performs the change subject table 32A (FIG. 9) of the personnel change data storage means 32. In the record of the transfer target person T7 (α), the department identification information of the determined transfer destination department (here, the department α as planned) is stored.

  Subsequently, the personnel change support means 24 proceeds to the arrangement process of T6 (α) and T4 (α) after finishing the arrangement process of T7 (α).

  In (Case 1) described above, the transfer target persons T6 (α) and T4 (α) do not have the empty frames α6 and α4 having the same cluster number (or cluster feature amount). Therefore, the personnel change support means 24 selects an empty frame whose cluster number (or cluster feature amount) is close to T6 (α) and T4 (α). As for T6 (α), α7 and α5 are close to each other, and either one may be selected. As for T4 (α), α5 and α3 are close to each other. However, since there is only one empty frame α5, it is not possible to make a random selection, and there are limited options. Therefore, the personnel transfer support means 24 selects the optimum option from the limited options. Select.

  Here, if T6 (α) is arranged in the empty frame α7 and T4 (α) is arranged in α5, the sum of the cluster numbers of the department α (or the sum of the cluster feature values) is reduced by two. End up. Further, if T6 (α) is arranged in the empty frame α5 and T4 (α) is arranged in α3, the sum of the cluster numbers of the department α (or the sum of the cluster feature values) increases by two. Therefore, the personnel change support means 24 selects to place T6 (α) in the empty frame α7 and to place T4 (α) in α3. Thus, the sum of the cluster numbers of the department α (or the sum of the cluster feature values) is maintained. Since the empty frames α5 and α2 are left over, in the subsequent processing (see step S6 described later), the transfer destinations whose transfer destination is undetermined are filled. If the sum of the cluster numbers (or the sum of the cluster feature values) of the department α has increased or decreased with the person to be transferred whose assignment destination has been determined to the department α being completed, the target of the transfer whose transfer destination is undetermined It adjusts at the time of a person's arrangement | positioning (refer step S6 mentioned later), or it adjusts by subsequent exchange (refer step S7 mentioned later).

  In addition to this, when the personnel change support means 24 is placed in an equivalent situation (a situation in which either cannot be determined) by the selection using the cluster number or the cluster feature amount, the Q1 selected by the user in step S1 described above. Detailed selection (secondary allocation selection) is performed by any one of methods Q4 (see FIG. 8).

  (Case 2) For example, it is assumed that the transfer target person is T5 (α), T2 (α), and the empty frames are α6, α4, α3, α1.

  At this time, there are two combinations of options for maintaining the sum of cluster numbers (or the sum of cluster feature values). The first option is to place T5 (α) at α6 and T2 (α) at α1. The second option is to place T5 (α) at α4 and T2 (α) at α3. Therefore, when the user selects Q1 in the above-described step S1 (see FIG. 8), the personnel change support means 24 determines the distance between T5 (α) and α6, and T2 ( The distance between α) and α1 is calculated, and the sum of these distances is obtained. For the second option, the distance between T5 (α) and α4 and the distance between T2 (α) and α3 are calculated, and the sum of these distances is obtained. Then, the personnel change support means 24 selects a combination option that reduces the sum of the distances from the possible combination options.

  Further, when the user selects Q2 in step S1 described above (see FIG. 8), the personnel change support means 24, for the first option, the difference in the number of customers in charge between T5 (α) and α6 , And the difference in the number of customers in charge between T2 (α) and α1, and the sum of these differences is obtained. For the second option, the difference in the number of customers in charge between T5 (α) and α4 and the difference in the number of customers in charge between T2 (α) and α3 are calculated, and the sum of these differences is obtained. Then, the personnel change support means 24 selects a combination option having a smaller difference sum among the possible combination options.

  Similarly, when the user selects Q3 in step S1 described above (see FIG. 8), the personnel change support means 24 uses the total value of the revenue amount instead of the above-mentioned number of customers in charge, and sets Q4. If selected, the cluster number (or cluster feature value) of the second classification process classified by the second cluster analysis based on the number of contacts is used instead of the number of customers in charge, and possible combinations of Among the choices, a combination choice that reduces the sum of differences is selected.

  For the transfer target person whose transfer destination is determined to be department α, the processing contents in the case where there is not enough free space in department α are as follows.

  (Case 3) For example, the transfer target persons who are determined to be transferred to the department α are T7 (α), T6 (α), and T4 (α), and the empty frames of the department α are α5 and α2. And Further, the department α is assumed to be a department that maintains the number of persons before and after the personnel change, that is, a department that does not allow the increase or decrease in the number of persons (increase or decrease excluding new employees). Therefore, since there are three sales employees on the transfer side to the department α and two sales persons on the transfer side from the department α, one additional sales person is transferred from the department α, and a vacant space is created. It must be increased by one.

  At this time, the sum of the cluster numbers of T7 (α), T6 (α), and T4 (α) is 17, and the sum of the cluster numbers of α5 and α2 is 7. Therefore, the difference for the sum of cluster numbers is 10. In order to reduce the change in the sum of the cluster numbers of department α (or the sum of cluster feature quantities) as much as possible before and after the personnel change, the cluster number of the transfer side sales staff and the transfer side sales staff Since the change in the sum (or the sum of the cluster feature values) must be reduced, the personnel change support means 24, when transferring an additional salesperson from the department α, from that perspective, Select the cluster number (or cluster feature) of the sales employee to be formed. However, the cluster number of the first classification process based on the first cluster analysis based on the revenue amount is only up to No. 7 in the example of FIG. 7, and is not No. 10, so that the sales staff of No. 7 is transferred to the vacant frame α7. Form. If there is no 7th salesperson, the 6th salesperson is transferred to form a vacant slot α6. If there is no 6th salesperson, the 5th salesperson is transferred to form a vacant slot α5. Perform processing such as. If, for example, the sales representative of No. 7 is transferred to form a vacant slot α7, the sales representative of No. 7 must be assigned to some department, so it becomes a transfer target person T7 (undecided) The personnel change support unit 24 creates a new record of the transfer target person T7 (undecided) in the transfer target person table 32A (see FIG. 9) of the personnel change data storage unit 32, and T7 (undecided) is added to the created record. Salesperson identification information, cluster number (or cluster feature amount), and information that the transfer destination is “undecided”.

  (State after addition of empty frame in case 3) For example, when the empty frame α7 is formed by transferring the 7th salesperson, the empty frames in (case 3) described above are 3 of α7, α5, and α2. Thus, the number of persons is balanced with the three transfer target persons T7 (α), T6 (α), and T4 (α).

  At this time, the personnel change support means 24 arranges T7 (α) in the empty frame α7 with the same cluster number (or cluster feature amount). T6 (α) and T4 (α) are arranged in α5 and α2 because there is no empty frame having the same cluster number (or cluster feature value). That is, T6 (α) is arranged in the empty frame α5, and T4 (α) is arranged in the empty frame α2. As a result, the sum of the cluster numbers of the department α increases by three, which is adjusted when the transfer destination is determined to be transferred (see step S6 to be described later) or exchanged thereafter ( Adjustment is made in step S7 described later).

  The personnel transfer support means 24 determines the transfer destinations to the departments β, γ, δ, ε,... Sequentially after the placement process of the transfer target person whose transfer destination is determined to the department α as described above. Then, after all the departments have been processed, the transfer target person whose transfer destination is undetermined is processed (step S6 in FIG. 8).

  Specifically, the personnel change support means 24 does not return the transfer target person T7 (undecided) whose transfer destination is undetermined to the vacant frames α7, β7, γ7, δ7, ε7,. Deploy. At this time, if the user selects Q1 in step S1 described above (see FIG. 8), the feature vector of T7 (undecided) and the feature vectors of the respective empty frames α7, β7, γ7, δ7, ε7,. And a detailed selection (secondary allocation selection) in which T7 (undecided) is placed in an empty frame with the shortest (close) distance to T7 (undecided) may be performed. Even if T7 (undecided) is placed evenly (in order) one by one in each department α, β, γ, δ, ε,. Well (in that case, 2 rounds, 3 rounds, 4 rounds,...) Or T7 (undecided) is randomly placed in the empty frames α7, β7, γ7, δ7, ε7,. May be.

  Further, the personnel change support means 24 is in charge of T7 (undecided) when the user selects Q2 in step S1 described above (see FIG. 8) when performing detailed selection (secondary allocation selection). Details of determining the difference between the number of customers and the number of customers in charge of each of the vacant frames α7, β7, γ7, δ7, ε7,... And placing T7 (undecided) in the vacant frame with the smallest difference from T7 (undecided) Make a selection. Similarly, when the user selects Q3 in step S1 described above (see FIG. 8), the total value of the revenue amount is used in place of the number of customers in charge, and when Q4 is selected, Use the cluster number (or cluster feature amount) of the second classification process classified by the second cluster analysis based on the number of contacts in place of the number of customers in charge, and the difference with T7 (undecided) is the smallest Detailed selection to arrange T7 (undecided) in the frame is performed.

  The personnel change support means 24 selects the transfer target person T7 (undecided) having the shortest (close) distance to the vacant frame α7 in the opposite way of thinking, and selects the selected T7 (undecided) as the vacant frame α7. Similarly, T7 (undecided) having the shortest (close) distance to the empty frame β7 is selected, the selected T7 (undecided) is arranged in the empty frame β7, and the distance to the empty frame γ7 is The shortest (close) T7 (undecided) may be selected, the selected T7 (undecided) may be placed in the empty frame γ7, and such processing may be performed for the empty frames δ7, ε7,.

  In addition, the personnel change support means 24 sets T7 (undecided) as empty when the number of empty frames α7, β7, γ7, δ7, ε7,... Is insufficient with respect to the number of transfer target persons T7 (undecided). Arranged in frames α6, β6, γ6, δ6, ε6,. On the other hand, if the empty frames α7, β7, γ7, δ7, ε7,... Remain after the placement of T7 (undecided), T6 (undetermined) is arranged there.

  Similarly, personnel change support means 24 arranges transfer target person T6 (undecided) whose transfer destination is undetermined in empty frames α6, β6, γ6, δ6, ε6,... So as not to return to the original department. To do. Then, if the number of empty frames α6, β6, γ6, δ6, ε6,... Is insufficient for the number of transfer target persons T6 (undecided), T6 (undecided) is changed to empty frames α5, β5, γ5, Arranged at δ5, ε5,. On the other hand, if the empty frames α6, β6, γ6, δ6, ε6,... Remain after the placement of T6 (undecided), T5 (undetermined) is arranged there.

  The personnel change support means 24 repeats such processing for the transfer target persons T5 (undecided), T4 (undetermined), T3 (undetermined), T2 (undetermined), and T1 (undetermined). Finally, empty frames are left for the number of retired persons, but normally, the empty frames α1, β1, γ1, δ1, ε1,. If there are a large number of retired employees, there may be remaining empty frames with cluster numbers such as 2 and 3.

  Also, the personnel change support means 24 performs the transfer target person table 32A (see FIG. 9) of the personnel change data storage means 32 each time the assignment target person whose transfer destination department is not determined as described above. ) Stores the department identification information of the determined transfer destination department, and stores the salesperson identification information of the arranged salesperson in the empty frame table 32B (see FIG. 10). This is the same as in the case of the transfer target person arrangement process in which the transfer destination department is determined.

  Thereafter, the personnel change support means 24 uses the data stored in the department table 32C (see FIG. 11) of the personnel change data storage means 32 and uses the data stored in the department table 32C (see FIG. 11) for each department, the cluster numbers before and after the personnel change. If the amount of increase / decrease of the sum of (or the sum of cluster features) is within a predetermined threshold, and if it is determined to exceed the predetermined threshold, Processing is performed so that the increase / decrease amount of the sum of the cluster numbers before and after the personnel change (or the sum of the cluster feature amounts) falls within the threshold (step S7 in FIG. 8).

  Note that the data stored in the department table 32C (see FIG. 11) is the personnel change support means each time a vacant frame is formed or a sales employee is arranged in the vacant frame in accordance with the personnel change. 24. In other words, when a vacant frame is formed, the same number as the cluster number (or cluster feature) of the salesperson who formed the vacant frame is reduced, and when a salesperson is placed in the vacant frame, the cluster number of the assigned salesperson A process of increasing the same quantity as (or cluster feature) is performed. The state shown in FIG. 11 is a state immediately after the formation of the vacant frame and is a state in which the sales staff is not yet placed in the vacant frame. The state illustrated in FIG. 11 corresponds to the empty frame formation state illustrated in FIG.

  Specifically, the personnel change support means 24 acquires the cluster number or the increase / decrease amount of the cluster feature before and after the personnel change stored in the department table 32C (see FIG. 11), and adds the positive department and the negative department. Then, exchange and adjust the sales staff. In the state illustrated in FIG. 11, there are only minus departments, but when the sales staff assignment process (see steps S5 and S6 in FIG. 8) is finished, a plus department and a minus department are generated. Become. At this time, salespeople who have already been transferred are excluded from replacement personnel. Further, information on which salesperson belongs to each department is obtained from the attribute table 30A (see FIG. 2) of the salesperson data storage means 30, and the cluster number of each salesperson as a candidate for replacement. (Or cluster feature amount) is acquired from the classification result storage means 31 (see FIG. 7). Note that if the salesperson replacement process does not fall within the predetermined threshold, the salesperson replacement process is performed a plurality of times.

  For example, in the case of a plus 3 department and a minus 3 department, the salesperson with the cluster number (or cluster feature) 7 is exchanged with the salesperson 4 or the salesperson 5 Perform processing to exchange with the 2nd salesperson. In the case of this example, the replacement of sales staffs 5 and 2 is the replacement of sales staff with a smaller cluster number (or cluster feature) than the replacement of sales staff 7 and 4. Therefore, it may be determined that the combination is substantially less affected by the exchange, and the exchange between the sales staffs No. 5 and No. 2 may be selected.

  Further, exchange may be performed between departments having different absolute values of plus and minus. For example, if the department is a plus 6 department and a minus 3 department, the salesperson with the cluster number (or cluster feature) 7 is exchanged for the salesperson 4 or the salesperson 4 If you exchange the number 1 salesperson, you will have a plus 3 department and a zero department. In this case, if the plus 3 department does not fall within the predetermined threshold, the sales staff may be exchanged with another department again.

  Each time the personnel change support means 24 performs the sales staff exchange process as described above, the personnel change support means 24 stores two persons related to the exchange in the change target person table 32A (see FIG. 9) of the personnel change data storage means 32. Create a new record for the sales representative (both are considered to be transferees), and store information including information after the exchange process (department identification information of the confirmed transfer destination department) in the created record, In the vacant frame table 32B (see FIG. 10), a new record is created for the two salespeople involved in the exchange (both are regarded as salespeople who have formed the vacant frame), and the created record is subjected to the exchange process. Information (salesperson identification information of the assigned salesperson) is stored.

  Since there are so many combinations of salesperson exchanges as described above, the personnel change support means 24 does not perform the detailed selection (secondary distribution selection), but performs the first cluster analysis based on the revenue amount. Random selection may be performed only by selection (first distribution selection) using the cluster number (or cluster feature amount) of the first classification processing according to.

  Finally, the output means 25 uses the data stored in each of the storage means 30 to 32 to display the sales staff classification result by cluster analysis and the result of the personnel change support process using the classification result. A process for displaying a screen on the screen is executed, and a series of processes relating to support for personnel changes is completed (step S8 in FIG. 8).

  <Effect of this embodiment>

  According to this embodiment, there are the following effects. In other words, in the salesperson clustering system 10, the feature vector creation means 22 creates, for each salesperson, a feature vector having each revenue amount of a plurality of quantiles in the revenue distribution for each salesperson as a constituent variable. Since the cluster analysis unit 23 performs cluster analysis using these feature vectors as input, it is possible to eliminate arbitraryness at the time of creating a cluster variable, and for the cluster that effectively captures the characteristics of a salesperson. Variables can be created, and the problems of the conventional methods that use aggregate values such as total values and average values as cluster variables can be solved.

  For example, with the conventional method, the impact of a case where a customer with a large profit amount can be acquired by chance (so-called lucky) during the aggregation target period may be reflected in the classification result. In the system 10, since the sales characteristics of the salesperson's individual can be grasped by the distribution by a plurality of quantiles, it is possible to realize a clustering that is not easily affected by chance or unique situations during the counting target period. it can.

  In other words, in the past, salesmen's quotas were often sales, and so on. In many cases, simple evaluations were made based on the sales amount. It is important that good business results are obtained), and since the present invention creates vector variables for clusters using multiple quantiles in the revenue distribution, such customer portfolio features A feature vector to be shown can be created, and stable classification can be performed. Because the conventional method makes a simple evaluation based on the sales amount, a salesperson who has one customer of 10 million yen and a salesperson who has 1000 customers of 10,000 yen They are classified into the same cluster. However, there are years when it is not possible to acquire 10 million yen customers, so the evaluation is not stable. On the other hand, since the salesperson clustering system 10 evaluates with the customer portfolio, the salespersons are not classified into the same cluster, and stable classification can be performed.

  In addition, by introducing the sales staff clustering system 10, organizations such as companies with a large number of sales staff can pattern individual sales characteristics of each sales staff, and sales strategies such as effective team organization. Can stand up.

  Furthermore, in the salesperson clustering system 10, since the cluster variable is a feature vector having the quantile profit amount as a constituent element, the same clustering is performed without changing the definition of the vector variable according to the application. can do.

  In addition, the salesperson clustering system 10 is configured to use a four-dimensional or higher feature vector having each quartile and higher quantile as the constituent elements, so that each quartile allows each A vector variable that captures the salesperson's individual profit characteristics evenly (no particular lack) and finely ticks the tail part (the bottom part of the large revenue amount) in the profit distribution by using the top quantile. Can be created. For this reason, according to Pareto's law (the majority of the overall characteristics are empirical rules created by some elements that make up the whole, also called the 80:20 law) Emphasized cluster analysis can be performed.

  Further, the feature vector creating means 22 is configured to determine the number and step size of the upper quantile according to the correspondence relationship between the predetermined customer number information and the number and step size of the upper quantile. Therefore, in accordance with Pareto's Law, when the tail of the revenue distribution is made finer, feature vectors can be created according to the number of customers in charge. A simple feature vector can be created. That is, when the number of customers in charge is large, the number of higher quantiles can be increased and the step size of the higher quantiles can be made narrower (smaller). For example, the top 0.1% quantile when the number of customers in charge is 1,000 indicates the revenue amount of the first customer from the top, but when the number of customers in charge is 400, the top 0.1% Since the profit amount of the percentile may become a meaningless value, such a situation can be avoided in advance.

  The feature vector creating means 22 can also create a feature vector having a maximum value, a total value, or an average value as an additional component in addition to each revenue amount of a plurality of quantiles. It is possible to perform cluster analysis that incorporates the concept of the conventional method in which the aggregate value such as the average value is used as a component of the feature vector.

  Furthermore, the salesperson clustering system 10 uses the classification result by the first cluster analysis using a plurality of quantiles in the distribution of revenue for each salesperson and the plurality of quantiles in the contact count distribution for each salesperson. Since the composite classification process combining the classification result by the second cluster analysis can be performed (when the user selects P2 in step S1 in FIG. 8), the individual revenue characteristics of each salesperson and each sales Clustering can be performed in combination with the individual customer service characteristics. For this reason, for example, it is possible to perform clustering that takes into account the personal characteristics of a salesperson, such as whether the salesperson has a large number of contacts with a specific customer or whether the salesperson contacts all customers equally. .

  Since the sales staff clustering system 10 includes the personnel change support means 24, when the sales staff is assigned to each department, the cluster numbers or cluster characteristics of the transfer-side sales staff and the transfer-side sales staff. A person selection process for automatically matching the quantities or a person selection process for matching the cluster numbers or cluster feature quantities of the two sales staff members when exchanging the sales staff members between the two departments. Can be executed automatically. For this reason, in each department, sales staff classified into the same cluster can be exchanged before and after personnel changes, and the balance of business processing capabilities between departments can be maintained.

  In addition, the personnel change support means 24 arranges sales persons (sales persons having the same cluster number or cluster feature amount) classified in the same cluster as sales persons in the empty frame, or classifies them in the same cluster. Under the principle of exchanging sales employees, when there are a plurality of candidates, it is possible to select the sales employee with the smallest distance between the two calculated using the feature vector (FIG. 8). When the user selects Q1 in step S1), the change in the business processing ability of each department α, β, γ,... Before and after the personnel change can be further reduced.

  Similarly, even when the user selects Q2 to Q4 in step S1 in FIG. 8, the change in the business processing ability of each department α, β, γ,... be able to. In particular, when Q4 is selected, the classification result based on the second cluster analysis based on the contact frequency distribution is used, so that detailed selection (secondary distribution selection) based on the individual customer service characteristics of each sales employee is performed. be able to. Therefore, in the case of performing a composite classification process that combines the classification result by the first cluster analysis based on the revenue distribution and the classification result by the second cluster analysis based on the contact frequency distribution (in step S1 in FIG. 8, the user selects P2 The classification result obtained by the second cluster analysis can be used in a different way from the case of selecting (1).

  <Deformation form>

  Note that the present invention is not limited to the above-described embodiment, and modifications and the like within a scope where the object of the present invention can be achieved are included in the present invention.

  For example, in the above-described embodiment, the second cluster analysis based on the number of contacts complements the first cluster analysis based on the amount of revenue, but in the case of a salesperson visiting a customer. May use, for example, the visit time instead of the contact count. In other words, the visit time for each salesperson and for each customer is stored in the salesperson data storage means 30 in association with the salesperson identification information and the customer identification information. Then, using the visit time data stored in the salesperson data storage means 30 by the feature vector creation means 22, the vertical axis is the number of customers and the horizontal axis is the visit time for one customer as a cluster variable. A process may be executed in which each visit time of a plurality of quantiles in the visit time distribution for each salesperson is obtained, and a feature vector having each found visit time as a constituent element is created for each salesperson.

  In the above embodiment, as shown in step S1 of FIG. 8, the selection input of the methods P1 and P2 of the first classification (primary distribution selection) to be classified into several clusters, and the arrangement of the sales staff However, the salesperson clustering system of the present invention uses each method as in the above-described embodiment. Instead of a configuration that can be freely selected, a system that is fixed in any one of the methods may be used.

  Furthermore, although the salesperson clustering system 10 of the said embodiment was set as the structure provided with the personnel transfer assistance means 24, the salesperson clustering system of this invention does not necessarily require the automation process of the operation | work which utilizes the classification result by cluster analysis. It is not necessary to use a system that includes, but automatically performs the clustering and outputs the classification results (information indicating which salesperson is classified into which cluster). You may make it perform the sales relocation work etc. of a sales employee manually using a classification result.

  As described above, the salesperson clustering system and program according to the present invention are used, for example, when supporting a task of reassigning a plurality of salespersons belonging to an organization such as a company composed of a plurality of departments by personnel change. Suitable for

DESCRIPTION OF SYMBOLS 10 Salesperson clustering system 22 Feature vector preparation means 23 Cluster analysis means 24 Personnel change support means 30 Salesperson data storage means 31 Classification result storage means

Claims (9)

A salesperson clustering system configured by a computer that executes a process of classifying a plurality of salespersons who make money by contacting customers,
Salesperson data storage means for storing the revenue amount for each salesperson and for each customer in association with the salesperson identification information and the customer identification information;
Using the revenue data stored in the salesperson data storage means, as a cluster variable, the vertical axis represents the number of customers, and the horizontal axis represents the revenue generated from one customer. A feature vector creating means for obtaining each revenue amount of a plurality of quantiles in the distribution, and executing a process of creating a feature vector having each found revenue amount as a constituent element for each salesperson;
Cluster analysis means for executing a process of classifying the plurality of sales staffs by cluster analysis with the feature vector for each sales staff created by the feature vector creating means as input,
The results of classification by this cluster analysis means and / or personnel transfer support processing using the classification results or other processing results are displayed on the screen, printed, transmitted to other systems via a communication line, and stored in a recording medium. Or a salesperson clustering system, comprising: an output means for executing a process for outputting the voice or sound. The feature vector creating means includes:
The three quartiles, the first quartile, the second quartile, and the third quartile, and the third quartile with the largest revenue among these three quartiles It is configured to execute a process of creating, for each salesperson, a four-dimensional or higher feature vector having a revenue amount of at least one higher quantile that is subdivided into a range in which the revenue amount is larger than a point. The salesperson clustering system according to claim 1. The feature vector creating means includes:
The number of customers in charge input by the user for the range, average number, maximum number or minimum number of customers in charge of one sales employee is used, or the data stored in the sales staff data storage means is used. And calculating the number of customers in charge, and the number and increments of upper quantiles to be subdivided according to the corresponding relationship between the number of customers in charge and the number of higher quantiles and the step size. The salesperson clustering system according to claim 2, wherein a process for determining a width is executed. The feature vector creating means includes:
In addition to each revenue amount of the plurality of quantiles, the maximum value of the revenue amount raised from each customer in charge of one sales employee and the revenue amount raised from each customer in charge of one sales employee Processing to create a feature vector for each salesperson that consists of at least one additional element selected from the total value of and the average value of the amount of revenue raised by each salesperson for each salesperson It is set as the structure to perform. The salesperson clustering system in any one of Claims 1-3 characterized by the above-mentioned. The salesperson data storage means includes
In addition to the amount of revenue for each salesperson and for each customer, the number of contacts for each salesperson and for each customer is stored in association with the salesperson identification information and customer identification information,
The feature vector creating means includes:
In addition to processing to create a feature vector for each salesperson with revenue amount as a component,
Using the contact count data stored in the salesperson data storage means, as a cluster variable, the vertical axis represents the number of customers, and the horizontal axis represents the number of contacts for each salesperson with the number of contacts to one customer. It is configured to execute the process of obtaining the number of each contact of a plurality of quantile points in, and creating a feature vector for each salesperson with each obtained contact number as a component,
The cluster analysis means includes
A first cluster analysis is performed to classify the plurality of sales employees into a first division number cluster by using a feature vector for each sales employee having the revenue amount created by the feature vector creation means as a component. In addition to the classification process
A second classification for classifying the plurality of sales employees into a second divided number of clusters by a second cluster analysis with the feature vector of each sales employee having the number of contacts created by the feature vector creation means as a component. Also perform processing,
Further, by combining the first classification result obtained in the first classification process and the second classification result obtained in the second classification process, the plurality of sales employees can be combined with the first classification result. The salesperson clustering system according to any one of claims 1 to 4, wherein the salesperson clustering system is configured to execute a composite classification process for classifying into a number of clusters obtained by multiplying the number of divisions and the second number of divisions. . The plurality of sales employees are sales employees belonging to an organization consisting of a plurality of departments,
The cluster analysis means includes
At least one type of cluster analysis is performed for all sales staff belonging to the organization, and each cluster obtained by one type of cluster analysis or a plurality of types of cluster analysis results is obtained. For each cluster obtained by the combined classification process, the cluster number assigned according to the level of excellence in sales results or the cluster feature amount indicating the level of excellence of each cluster is assigned to the sales staff classified into each cluster. It is configured to execute a process of storing in the classification result storage means in association with the salesperson identification information,
With personnel change support means to execute personnel change support processing that supports the relocation work between sales staff departments,
This personnel change support means
When placing a sales employee in each department, a cluster number or a cluster feature stored in the classification result storage means in association with sales representative identification information about the transfer-side sales employee placed in each department, A cluster number or a cluster feature stored in the classification result storage means in association with the salesperson identification information about the salesperson on the transfer side who has formed a vacancy in each department due to a change from department to other department. When the matching person selection process is executed, or when salespeople are exchanged between two departments, they are stored in the classification result storage means in association with the salesperson identification information about the two salespeople to be exchanged. The salesperson class according to any one of claims 1 to 5, wherein a person selection process for matching a cluster number or a cluster feature amount is executed. Stalling system. The salesperson data storage means includes
The feature vector of each salesperson created by the feature vector creation means is also configured to be stored in association with salesperson identification information,
The personnel change support means is:
When assigning sales employees to each department, if there are multiple sales side salespersons whose cluster numbers or cluster feature values match the cluster number or cluster feature value of the transfer side salesperson, or move out When there are multiple transfer-side salespeople whose cluster numbers or cluster feature values match the cluster number or cluster feature amount of the salesperson on the side, or when exchanging salespeople between two departments, When there are a plurality of replacement salespeople with the same cluster number or cluster feature quantity, the distance between the two calculated by using the feature vector stored in the salesperson data storage means is the smallest. The salesperson clustering system according to claim 6, wherein the salesperson clustering system is configured to execute a process of selecting a salesperson. The salesperson data storage means includes
In addition to the amount of revenue for each salesperson and for each customer, the number of contacts for each salesperson and for each customer is stored in association with the salesperson identification information and customer identification information,
The feature vector creating means includes:
In addition to processing to create a feature vector for each salesperson with revenue amount as a component,
Using the contact count data stored in the salesperson data storage means, as a cluster variable, the vertical axis represents the number of customers, and the horizontal axis represents the number of contacts for each salesperson with the number of contacts to one customer. It is configured to execute the process of obtaining the number of each contact of a plurality of quantile points in, and creating a feature vector for each salesperson with each obtained contact number as a component,
The plurality of sales employees are sales employees belonging to an organization consisting of a plurality of departments,
The cluster analysis means includes
First, the plurality of sales employees are identified by the first cluster analysis with the feature vector of each sales employee having the revenue amount created by the feature vector creation means as a component for all sales employees belonging to the group. In addition to the first classification process for classifying the cluster into 1 division number,
A feature vector for each salesperson having the number of contacts created by the feature vector creation means as a component for all salespersons belonging to the group as an input is input to the plurality of salespersons by the second cluster analysis. A second classification process for classifying the cluster into a number of division clusters,
Furthermore, a cluster number given according to the degree of excellence of the business results for each cluster obtained in the first classification process or a cluster feature amount indicating the degree of excellence of each cluster is obtained in the first classification process. The classification result storage means stores the information in association with the sales staff identification information for the sales staff classified into each obtained cluster, and the degree of excellence in sales performance for each cluster obtained in the second classification processing. The classification result obtained by associating the cluster number assigned according to the above or the cluster feature amount indicating the degree of excellence of each cluster with the salesperson identification information for the salesperson classified into each cluster obtained in the second classification processing It is configured to store in the storage means,
With personnel change support means to execute personnel change support processing that supports the relocation work between sales staff departments,
This personnel change support means
The cluster number of the first classification process stored in the classification result storage means in association with the salesperson identification information about the transfer-side salesperson arranged in each department when the salesperson is arranged in each department Alternatively, the cluster feature value is stored in the classification result storage means in association with the salesperson identification information about the salesperson on the transfer side who has formed an empty frame in each department due to the transfer from each department to another department. Salesperson identification of two salespersons to be exchanged when executing a personnel selection process to match the cluster number or cluster feature amount of the first classification process or exchanging salespersons between two departments Performing a person selection process that matches the cluster number or cluster feature amount of the first classification process stored in the classification result storage means in association with information;
Further, when a salesperson is assigned to each department, the cluster number or cluster feature value of the first classification process is different from the cluster number or cluster feature value of the first classification process for the salesperson on the transfer side. When there are a plurality of transfer side salespersons that match or the cluster number or cluster feature value of the first classification process for the transfer side salesperson, the cluster number or cluster feature of the first classification process When there are a plurality of transfer side salespeople with the same amount, or when exchanging salespeople between two departments, a plurality of replacement objects with the same cluster number or cluster feature amount of the first classification process If there is a sales employee, the cluster number or the cluster feature amount of the second classification process matches or a process for selecting a sales person close to is executed. Salespeople clustering system according to any one of claims 1 to 4, characterized in that there is a formed.   The program for functioning a computer as a salesperson clustering system in any one of Claims 1-8.

Images