Retirement risk assessment system and retirement risk assessment method

Abstract

Even when data on former employees is limited, or when the characteristics of employees identified as having a high risk of leaving by the employee turnover risk assessment model are unclear, this system makes it possible to identify employees with a high risk of leaving and analyze the factors that led to that assessment. [Solution] The employee resignation risk assessment device 103 communicates with a virtual PC 108 and a management server 107 using a communication device 2002. The device acquires operation history from the virtual PC 108, resource consumption history from the management server 107, and personal data of the employee to be assessed. Based on the operation history, resource consumption history, and employee personal data, the device performs machine learning on the employee resignation risk assessment model 102 to determine the employee's resignation risk.

Description

Technical Field

[0001] The present invention relates to a retirement risk determination system and a retirement risk determination method for determining the retirement risk of workers.

Background Art

[0002] In recent years in Japan, due to the decline of lifetime employment and changes in awareness towards job-hopping, the mobility of human resources has been increasing, and the outflow of human resources due to retirement has become an issue in management. Therefore, many companies are seeking to more accurately identify the reserve army of retirees and avoid the outflow of human resources by analyzing employees' personal data and the operation history of PCs (personal computers).

[0003] In Patent Document 1, based on the employee's registration management information, the operation history and resource consumption history of the PC used by the employee are classified as being by a retired employee or an in-service employee to create learning data. And it is described that machine learning of a retirement risk determination model is executed using the learning data, and the retirement risk is determined using the retirement risk determination model.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, the determination method described in Patent Document 1 has the following problems. First, the data of retired employees that can be obtained annually will be scarce. Also, even if a large amount of data of retired employees from the past is prepared, it is not data corresponding to changes in the times such as changes in work styles and the use of job-hopping sites, and there is a problem that it is difficult to expect the accuracy of the data of retired employees.

[0006] Furthermore, it is unclear what factors were used to determine that an employee is at high risk of leaving using the employee turnover risk assessment model. This presents a challenge in creating an environment where follow-up can be carried out quickly and accurately for employees who are identified as being at high risk of leaving.

[0007] In view of the above problems, the present invention aims to identify employees at high risk of resignation and to analyze what factors led to this determination, even when there is limited data on former employees or when the characteristics of employees identified as having a high resignation risk by a resignation risk assessment model are unknown. [Means for solving the problem]

[0008] To achieve the above objective, an employee turnover risk determination system according to one aspect of the present invention is characterized by comprising: a storage device that holds personal data of an employee; a communication device that communicates with a virtual PC used by the employee for work and a management server for the virtual PC; and a determination device that, via the communication means, acquires the employee's operation history from the virtual PC and the resource consumption history of the virtual PC from the management server, performs machine learning of an employee turnover risk determination model based on the employee's personal data, the operation history, and the resource consumption history, and determines the employee's turnover risk.

[0009] Furthermore, a method for determining employee turnover risk according to one aspect of the present invention is characterized by comprising the steps of: viewing personal data of an employee; obtaining the employee's operation history and the virtual PC's resource consumption history from the virtual PC used by the employee for work and the management server of the virtual PC; and performing machine learning of a turnover risk determination model based on the employee's personal data, the operation history, and the resource consumption history to determine the employee's turnover risk. [Effects of the Invention]

[0010] According to the present invention, even when there is limited data on retired employees, or when the characteristics of employees identified as having a high risk of retirement by a retirement risk assessment model are unknown, it is possible to identify employees with a high risk of retirement and analyze what factors led to this assessment. [Brief explanation of the drawing]

[0011] [Figure 1] Diagram showing the retirement risk assessment system of this embodiment. [Figure 2] This figure shows an example of the main flow of the retirement risk assessment method in this embodiment. [Figure 3] A diagram showing an example of the data collection flow in this embodiment. [Figure 4] This figure shows an example of the configuration of a virtual PC in this embodiment. [Figure 5] This figure shows an example of the operation history of the virtual PC in this embodiment. [Figure 6] This figure shows an example of the upload timing from the virtual PC in this embodiment. [Figure 7] This figure shows an example configuration of the management server for the virtual PC in this embodiment. [Figure 8] Figure showing an example of performance data in this embodiment. [Figure 9] This figure shows an example configuration of the data collection server in this embodiment. [Figure 10] This figure shows an example of the data type classification flow in this embodiment. [Figure 11] This figure shows an example of the configuration of the connection information table in this embodiment. [Figure 12] This figure shows an example of the configuration of the file operation information table in this embodiment. [Figure 13] This figure shows an example of the configuration of the site browsing information table in this embodiment. [Figure 14] This figure shows an example of the configuration of the performance data table in this embodiment. [Figure 15] This figure shows an example of the configuration of the personal data table in this embodiment. [Figure 16]Figure showing an example of the data aggregation flow in this embodiment [Figure 17] Figure showing an example of the monthly data aggregation table in this embodiment [Figure 18] Figure showing an example of the list of explanatory variables in this embodiment [Figure 19] Figure showing a configuration example of the risk determination server in this embodiment [Figure 20] Figure showing a configuration example of the retirement risk output result table in this embodiment [Figure 21] Figure showing a configuration example of the principal component analysis output result table in this embodiment [Figure 22] Figure showing an example of the flow of the behavior characteristic analysis of retirees in this embodiment [Figure 23] Figure showing a configuration example of the exception reason table in this embodiment [Figure 24] Figure showing an example of a table that summarizes various data in monthly format for employees with a high retirement risk in this embodiment

Embodiments for Carrying Out the Invention

[0012] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. Note that the following embodiments do not limit the invention described in the claims. Also, although a plurality of technical features are described in the following embodiments, not all of these plurality of features are essential to the present invention and may be used arbitrarily. Furthermore, in the drawings, the same or similar configurations are given the same reference numerals, and duplicate explanations are omitted.

[0013] Figure 1 is a diagram illustrating an example of a network configuration for explaining the employee turnover risk assessment system 100 according to this embodiment. The employee turnover risk assessment system 100 shown in Figure 1 efficiently collects data to be processed via the network 101. This computer system can identify abnormal data, i.e., employees with a high turnover risk, and analyze their characteristics, even when there is little data on retired employees in response to recent changes in the times, or when it is unclear what factors led to an employee being determined to have a high turnover risk using conventional methods.

[0014] As shown in Figure 1, the retirement risk assessment system 100 according to this embodiment is configured such that a risk assessment server 103, an employee terminal 104, an external assessment device 105, a data collection server 106, a virtual PC 108 used by employees for work, and a management server 107 for the virtual PC 108 are all connected via a network 101 in a manner that allows for communication between them.

[0015] The risk assessment server 103 (retirement risk assessment device), which mainly constitutes the retirement risk assessment system 100 of this embodiment, works in cooperation with the management server 107 and data collection server 106 (and external assessment device 105 if necessary) of the virtual PC 108 to acquire operation history and performance data (resource consumption history) of the virtual PC at high frequency and as appropriate, and performs retirement risk assessment. The retirement risk assessment is performed by a retirement risk assessment model 102 (details described later) which is generated by machine learning and tuned as appropriate.

[0016] Therefore, it can be assumed that the retirement risk assessment system 100 is configured by combining the risk assessment server 103 with at least one of the following: an external assessment device 105, a data collection server 106, and a management server 107 for the virtual PC 108.

[0017] The virtual PC 108 is virtually configured based on appropriate hardware resources (which may be provided by the management server 107 or by another appropriate system) and is made available for use via the network 101.

[0018] Furthermore, employee terminals 104 are terminals used by each employee of the company, etc., when using the virtual PC 108 described above. These employees operate employee terminals 104 to access the management server 107 via the network 101, and after logging in to the management server 107, they operate the desktop of the virtual PC 108. Specific examples of employee terminals 104 include personal computers, tablet devices, and smartphones.

[0019] Furthermore, the management server 107 for the virtual PC 108 is a server device that performs appropriate control, such as access management and scalability management, related to the virtual PC 108 as described above. This management server 107 monitors the resource consumption of each virtual PC 108 and distributes the performance data, which is the data collected by each virtual PC 108, to the data collection server 106.

[0020] Furthermore, the data collection server 106 is a server device that acquires the operation history of employees on the virtual PC 108, which is distributed in real time or at predetermined intervals at a high frequency from the virtual PC 108 (agent), and provides this to the risk assessment server 103. The data collection server 106 also distributes the performance data obtained from the management server 107 mentioned above to the risk assessment server 103.

[0021] Furthermore, the external determination device 105 is a device that performs retirement risk determination using a different logic than the risk determination server 103. The logic executed by this external determination device 105 can be appropriately assumed to be, for example, one that is already on the market using existing technology.

[0022] The external determination device 105, in the process of determining employee turnover risk in the risk determination server 103, acquires data on the employee being determined from the risk determination server 103, the data collection server 106, or the management server 107, or other appropriate devices, and performs an employee turnover risk determination based on that data. The result of this determination is provided to the risk determination server 103 and merged with the employee turnover risk determination result in the risk determination server 103 (of course, this is not mandatory).

[0023] <Method for Assessing Resignation Risk> Next, the procedure for determining employee turnover risk in this embodiment will be explained with reference to the diagram. The various operations corresponding to the employee turnover risk determination method described below are implemented by programs that are mainly read into memory and executed by the risk determination server 103 and data collection server 106, which constitute the employee turnover risk determination system 100. These programs consist of code for performing the various operations described below.

[0024] Figure 2 shows an example of the main flow of the retirement risk determination method in this embodiment. First, the virtual PC 108 and the data collection server 106 work together to collect the operation history and performance data of the virtual PC 108 (S201).

[0025] The details will be explained based on Figures 3 to 8. Figure 3 shows an example of the data collection flow in this embodiment. Each step of this data collection is led and controlled by the virtual PC 108. The breakdown of the process is as shown in Figure 3, consisting of operation history data collection S2011 and performance data collection S2012.

[0026] The configuration example of the virtual PC 108 is assumed to be as shown in Figure 4. Figure 4 is a diagram showing an example of the configuration of the virtual PC in this embodiment. The virtual PC 108 in this embodiment has a CPU (Central Processing Unit) 401, a communication device 402, memory 403, and a storage device 404.

[0027] The memory 403 includes a connection-related data collection function 405 that collects data on login time, logoff time, and source IP address; an access destination collection function 406 that collects the value of accessed URLs; a file operation collection function 407 that collects data on opened files; and a data upload function 408 that sends the operation history obtained by each of these collection functions to the risk assessment server 103. These functions are implemented by the CPU 401 executing corresponding programs, for example, stored in the storage device 404. The operation history 409 obtained by the above-mentioned collection functions is stored and retained in the storage device 404 for a certain period of time, for example.

[0028] A specific example of the operation history 409 described above is shown in Figure 5. Figure 5 is a diagram showing an example of the operation history of a virtual PC in this embodiment. The operation history 409 shown here associates the date and time obtained when the operation history was observed with values ​​such as the operation performed at that date and time, and the type of operation. The values ​​in the operation column include data related to the login time, logoff time, and source IP address of the virtual PC 108, the accessed URL, and the opened file name.

[0029] Furthermore, it is preferable that the data upload function 408 of the virtual PC 108 uploads the operation history 409 (i.e., sends data to the risk assessment server 103) at the timing indicated by the instruction from the data collection timing control function 9031 (described later) of the data collection server 106.

[0030] Figure 6 illustrates the concept of distributed uploads by each virtual PC 108 based on such upload timings. Figure 6 shows an example of upload timings from virtual PCs in this embodiment. In the example in Figure 6, it is assumed that there is a table 601 that defines the corresponding upload schedule for each virtual PC 108, for each virtual PC 108 identification information (e.g., a fixed IP address or the last digit of a MAC address). This table 601 is assumed to be maintained and available to the data collection timing control function 9031 of the data collection server 106 described above. In this embodiment, the upload timing is distributed for each virtual PC as shown in the example upload schedule in Figure 6.

[0031] Next, we will explain the procedure for collecting performance data. The management server 107 of the virtual PC 108 will be responsible for collecting the performance data. This management server 107 will upload the performance data collected for each virtual PC 108 to the risk assessment server 103.

[0032] Figure 7 shows an example configuration of the management server 107 in this embodiment. Figure 7 is a diagram showing an example configuration of the management server for the virtual PC in this embodiment. The management server 107 in this embodiment is a server device that manages resources (such as physical hardware and its control applications) for implementing the virtual PC 108.

[0033] This management server 107 is equipped with a CPU 701, a communication device 702, memory 703, and storage device 704.

[0034] Of these, the storage device 704 is composed of non-volatile storage means such as an HDD (Hard Disk Drive) or SSD (Solid State Drive). The memory 703 is composed of volatile storage means such as RAM (Random Access Memory). This memory 703 is equipped with a resource information collection function 705, which collects performance data 706 from each virtual PC 108 and stores it in the storage device 704. The CPU 701 is a processing unit that executes various processes on the management server 107 and implements the necessary functions. The communication device 702 is assumed to be a network interface card that accesses the network 101 and communicates with other devices connected to the network 101.

[0035] Figure 8 illustrates a specific example of the performance data 706 stored in the storage device 704 and uploaded from the management server 107 to the risk assessment server 103. Figure 8 is a diagram showing an example of performance data in this embodiment.

[0036] As shown here, performance data 706 associates the date and time the performance data was acquired with values ​​such as the CPU usage, network traffic (receiving; inbound), and network traffic (sending; outbound) for each virtual PC 108.

[0037] Now, let's return to the explanation of the main flow (Figure 2). The data acquisition server 106 performs data classification (S202) on the operation history and performance data of the virtual PC 108 obtained by S201 described above.

[0038] First, an example configuration of the data collection server 106 that executes S202 will be explained based on Figure 9. Figure 9 is a diagram showing an example configuration of the data collection server in this embodiment. The data collection server 106 has a CPU 901, a communication device 902, a memory 903, and a storage device 904. The storage device 904 is composed of non-volatile storage means such as an HDD or SSD. The memory 903 is composed of volatile storage means such as RAM. This memory 903 is equipped with a data collection timing control function 9031, a personal data viewing function 9032, a data type classification function 9033, a data aggregation function 9034, and an aggregated data upload function 9035, and classifies and aggregates personal data obtained from the operation history 409, performance data 706, and the personal data viewing function 9032, and stores the results in the storage device 904. It also uploads the results to the risk assessment server 103 at an appropriate time.

[0039] Of the functions described above, the data acquisition timing control function 9031, as already mentioned, is a function that schedules the upload timing of the operation history 409 from the virtual PC 108 to the data acquisition server 106 to be distributed among the respective virtual PCs 108, and notifies the corresponding virtual PC 108.

[0040] Furthermore, personal data viewing function 9032 is a function that requests and obtains various personal data of employees (e.g., attendance records, paid leave records, job title, etc.) from the human resources management system, etc.

[0041] Furthermore, the data type classification function 9033 is a function described later using Figure 10, and is a function that classifies each piece of data, including operation history 409 obtained from the virtual PC 108 and various personal data obtained from the personal data viewing function 9032, into categories.

[0042] Furthermore, the data aggregation function 9034 performs the data aggregation process shown in Figure 16 and later on the data classified by the data type classification function 9033. Details will be described later.

[0043] Furthermore, the aggregated data upload function 9035 uploads the aggregated results, which are the processing results of the data aggregation function 9034, to the risk assessment server 103.

[0044] Furthermore, the CPU 901 is an arithmetic unit that executes various processes in the data acquisition server 106 and implements the necessary functions.

[0045] Furthermore, the communication device 902 is assumed to be a network interface card that accesses network 101 and communicates with other devices on network 101.

[0046] Next, the detailed flow of S202 in the data collection server 106 will be explained based on Figure 10. Figure 10 is a diagram showing an example of the data type classification flow in this embodiment. In the flow shown in Figure 10, the data collection server 106 performs connected data type classification S2021.

[0047] This connection data type classification is a process that extracts the operation history 409 that contains URL values ​​and generates the connection information table 9041 shown in Figure 11. Figure 11 is a diagram showing an example of the configuration of the connection information table. In this process, the data collection server 106, for example, refers to a pre-held internal IP address range table 9048 to determine which location of the company the "connection source" is.

[0048] Next, the data collection server 106 performs S2022, which classifies the application used and the type of file opened.

[0049] S2022 is a process in which the data collection server 106 extracts file operation-related information from the operation history 409 and generates the file operation information table 9042 shown in Figure 12. Figure 12 is a diagram showing an example of the configuration of the file operation information table. In this process, the data collection server 106, for example, refers to a pre-held file type classification table 9049 and makes a classification determination according to the string contained in the file name.

[0050] Next, the data collection server 106 performs the classification S2023 of the type of website that was viewed.

[0051] This classification of the type of website visited is a process in which the data collection server 106 extracts information related to the URLs of accessed websites from the operation history 409 and generates the site browsing information table 9043 shown in Figure 13. Figure 13 is a diagram showing an example of the structure of the site browsing information table. In this process, the data collection server 106, for example, refers to a pre-held URL domain and website type classification table 9050 and determines the type of website according to the URL domain.

[0052] Next, the data collection server 106 executes S2024, which calculates the monthly average CPU usage rate for each employee.

[0053] The calculation of the monthly average CPU usage rate for each employee is a process in which the data collection server 106 extracts the CPU usage rate and network traffic values ​​shown in the performance data 706 to generate the performance data table 9044 shown in Figure 14. Figure 14 shows an example of the configuration of the performance data table.

[0054] Next, the data collection server 106 performs the classification of personal data type S2025 and completes the flow shown in Figure 10.

[0055] The classification of personal data is a process in which the data collection server 106 extracts various values, such as attendance records, paid leave records, job titles, and departmental affiliations, as indicated by the personal data viewing function 9032, and generates the personal data table 9045 shown in Figure 15. Figure 15 is a diagram showing an example of the configuration of the personal data table. In this process, the data collection server 106 refers to a personal data classification table 9051, which is held in advance and contains information such as leave type and job title, and determines the classification according to the personal data.

[0056] Now, let's return to the explanation of the main flow (Figure 2). The data collection server 106 performs data aggregation on the operation history and performance data of the virtual PC 108, as well as personal data, which were classified up to S202 as described above (S203).

[0057] The data collection server 106, which executes S203, then performs the monthly data aggregation S3011 shown in Figure 16. Figure 16 is a diagram showing an example of the data aggregation flow in this embodiment. In S3011, the data collection server 106 extracts the login time and logoff time of each employee's virtual PC 108, as well as the time when the employee is performing work (performing some operation on the subject of work) even outside of those hours, based on the date and time data in each record of the connection information table 9041, the opened file information table 9042, the site browsing information table 9043 showing the websites visited, the performance data table 9044, and the personal data table 9045 that have been generated up to this point. The data is then aggregated for each employee for predetermined periods such as the most recent month, one month ago, two months ago, and three months ago. Figure 17 is a diagram showing an example of a monthly data aggregation table. The data collection server 106 generates a monthly data summary table 9046 based on the aggregation in S3011, recording the most recent total monthly working hours, the total monthly working hours one month prior, the total monthly working hours two months prior, the total monthly working hours three months prior, etc., for each employee ID.

[0058] Next, the data collection server 106 performs data deletion S3012, which removes data unnecessary for analysis, and terminates the flow shown in Figure 16. In S3012, the data collection server 106 deletes data unnecessary for the employee turnover risk assessment model 102 (such as year and month, employee ID, and monthly behavioral data), and refers to the list of explanatory variables exemplified in Figure 18 to use the data shown in 1801 for machine learning.

[0059] Now, let's return to the explanation of the main flow (Figure 2). The risk assessment server 103 performs machine learning on the retirement risk assessment model 102 based on the aggregated results obtained from the processing up to S203 by the data collection server 106 described above (S204). In the machine learning performed in S204, for example, a non-hierarchical clustering method is used, and the number of clusters is specified as two.

[0060] The risk assessment server 103 also performs an employee turnover risk assessment and outputs the assessment results to the employee turnover risk output result table 2007, which will be described later (S205).

[0061] Figure 19 shows an example of the configuration of a risk assessment server. In this embodiment, the risk assessment server 103 has a configuration that includes a CPU 2001, a communication device 2002, a memory 2003, and a storage device 2004, as illustrated in Figure 19. Of these, the storage device 2004 is composed of a non-volatile storage means such as an HDD or an SSD.

[0062] The employee turnover risk assessment function 2005 is a function that determines the employee's turnover risk by inputting the employee's operation history 409, performance data 706, and various personal data, which are appropriately classified and aggregated, into the employee turnover risk assessment model 102. The result of this assessment is the employee turnover risk output result table 2007 (Figure 20), which outputs the employee ID, the data used for machine learning (Figure 18), and the clustering judgment result from the employee turnover risk assessment function 2005 described above. Figure 20 shows an example of the structure of the employee turnover risk output result table. As shown in this table 2007, a table linked to the employee turnover risk assessment result for each employee is output, using the employee ID as the key.

[0063] Furthermore, the principal component analysis function 2006 in Figure 19 refers to the judgment results from the retirement risk judgment function 2005 mentioned above, and the employment information contained in the personal data table, and performs principal component analysis on the data of employees classified into clusters that include many employees who are no longer employed. Figure 21 shows an example of the configuration of the principal component analysis output result table. The principal component analysis output table 2008 shown in Figure 21 is obtained by the analysis using the principal component analysis function 2006, and outputs eigenvalues, contribution rates, cumulative contribution rates, and each principal component loading.

[0064] Furthermore, the CPU 2001 in Figure 19 is a processing unit that executes various processes in the risk assessment server 103 and implements the necessary functions. The communication device 2002 is assumed to be a network interface card that accesses the network 101 and communicates with other devices on the network 101.

[0065] Now, let's return to the explanation of the main flow (Figure 2). The risk assessment server 103 performs an analysis of the behavioral characteristics of employees who have left the company (S206), based on the operation history 409 and performance data 706 of each employee, various personal data, aggregated data obtained by aggregating these, and the results of the assessment of the resignation risk assessment model 102, as described above, up to S205.

[0066] Figure 22 shows an example of a flow chart for analyzing the behavioral characteristics of former employees. In S206, the risk assessment server 103 refers to the assessment results from the former employee risk assessment function 2005 and the personal data table, and performs principal component analysis on the data of employees classified into clusters that include many employees who are no longer employed (S2061).

[0067] In this principal component analysis, the risk assessment server 103 outputs a principal component analysis output table 2008 (Figure 21) as the assessment result, which includes eigenvalues, contribution rates, cumulative contribution rates, and principal component loadings. By referring to the eigenvalues ​​and contribution rates, the server determines how many principal components to use for the analysis. For example, if the server decides to use principal components with eigenvalues ​​of 1 or more and cumulative contribution rates of 80% or more, it will use the first and second principal components from the principal component analysis output table 2008 for the analysis. Furthermore, each element of the principal component loadings is the same as the elements of the data used for machine learning (Figure 18), indicating how much each element contributes to each principal component.

[0068] Next, the risk assessment server 103 performs an analysis to determine whether the assessment is appropriate (S2062), and terminates the flow shown in Figure 22. Figure 23 shows an example of the structure of the exception reason table. In the analysis in S2062, the risk assessment server 103 determines the elements with the greatest impact from the principal component analysis output table 2008. If there is a reason to exclude the employee who underwent principal component analysis for the above-mentioned elements with the greatest impact, it refers to the exception table 2009 and excludes the employee if the threshold is exceeded. The exception table 2009 should be created as appropriate by determining the threshold and exception reasons for each company, department, or job.

[0069] Figure 24 shows an example of a table summarizing various data for employees at high risk of leaving the company in a monthly format. For employees who have been determined to be at high risk of leaving the company after the exclusion process using S2062, the various data that were processed or obtained as a result of the processing in each step up to this point are compiled into a monthly table 2010 as shown in Figure 24, and this data is distributed to employee terminals 104, such as the employee's supervisor or HR manager.

[0070] According to this embodiment, it becomes possible to present to HR personnel and others what kind of operation history is linked to the risk of employee turnover, and it is expected that this will lead to the creation of an environment in which follow-up with the employee in question can be carried out quickly and accurately. In addition, it becomes possible to efficiently collect the data to be processed and to perform employee turnover risk assessment with more appropriate accuracy.

[0071] The numerical values, processing timing, processing order, processing entity, data (information) acquisition method / destination / source / storage location, etc., used in the above embodiment are given as examples for the purpose of providing a concrete explanation, and are not intended to limit the scope to such examples.

[0072] Furthermore, some or all of the embodiments described above may be used in appropriate combinations. Alternatively, some or all of the embodiments described above may be used selectively.

[0073] The invention is not limited to the embodiments described above, and various modifications and changes are possible within the scope of the gist of the invention. [Explanation of symbols]

[0074] 101 Network 102 Retirement Risk Assessment Model 103 Risk Assessment Server 104 Employee terminals 105 External judgment device 106 Data Acquisition Server 107 Virtual PC Management Server 108 Virtual PCs

Claims

1. A storage device that holds personal data about employees, The aforementioned employee uses a virtual PC for work and a communication device that communicates with the management server of the said virtual PC, A determination device that, via the communication device, acquires the employee's operation history from the virtual PC, acquires the virtual PC's resource consumption history from the management server, performs machine learning on an employee turnover risk determination model based on the employee's personal data, the operation history, and the resource consumption history, and determines the employee's turnover risk. A retirement risk assessment system characterized by having the following features.

2. The retirement risk determination system according to claim 1, characterized in that the determination device performs principal component analysis based on the results of the machine learning, outputs elements that have a significant impact on the determination of retirement risk, and if there is a reason to exclude the elements with a significant impact, it excludes them by referring to an exception table.

3. The determination device refers to the employee's employment information stored in the storage device along with the personal data, The data input to the aforementioned retirement risk assessment model is unclassified data for both retired and current employees, The employee turnover risk assessment system according to claim 1 is characterized in that the employee turnover risk assessment model performs principal component analysis on data of employees classified into clusters that include a large number of employees who are no longer employed.

4. The retirement risk determination system according to claim 1, characterized in that the determination device acquires data as operation history, which includes at least one of the following: login and logoff times on the virtual PC, the source IP address, the file being operated on, and the URL to which it accessed, and acquires CPU usage data as resource consumption history.

5. The retirement risk determination system according to claim 1, further comprising a data collection server connected to the communication device, which acquires the operation history from the virtual PC in real time or at predetermined high-frequency intervals, acquires the resource consumption history from the management server, and provides the operation history and the resource consumption history to the determination device.

6. The retirement risk determination system according to claim 1, characterized in that the determination device performs principal component analysis based on the results of the machine learning, outputs the elements that have a large influence on the determination of retirement risk, and if there is a reason to exclude the elements with a large influence, it excludes them by referring to an exception table.

7. The retirement risk determination system according to any one of claims 1 to 6, characterized in that the determination device outputs the determined retirement risk for each employee in a monthly table format.

8. The process of viewing personal data about employees, The process of obtaining the employee's operation history and the virtual PC's resource consumption history from the virtual PC used by the employee for work and the management server of the virtual PC, A method for determining employee turnover risk, comprising the steps of: performing machine learning on an employee turnover risk determination model based on the employee's personal data, the operation history, and the resource consumption history; and determining the employee's turnover risk.

Images