Support tool for analysis related to human capital management

Abstract

Providing technical means to automate and support advanced analysis related to the optimization of human capital management on a distributed platform where information is securely managed. [Solution] The human capital management analysis support device 1 according to the present invention comprises a personnel information acquisition unit 111, a demographic data acquisition unit 112, a prediction unit 113 that generates prediction results of the trend in the number of employees by probabilistic trials using a probability model based on personnel information, and an output unit 116 that outputs an average value and confidence interval for the trend calculated by statistical processing of a plurality of the prediction results. The prediction unit 113 uses a probability model that includes at least a first parameter related to the number of hires and a second parameter related to the number of retirees, and generates prediction results using the first parameter corrected by future projections.

Description

[Technical Field]

[0001] This invention relates to a support device for analysis related to human capital management. [Background technology]

[0002] The concept of human capital management, which optimizes employee allocation within a company, is gaining attention. In human capital management, allocation is optimized based on employee skills, place of residence, age, gender, and other attributes. In other words, human capital management uses employee data and other data corresponding to the aforementioned attributes to perform optimization.

[0003] However, such data is personal information and may contain trade secrets, requiring strict management in its handling. Therefore, securing personnel capable of strictly managing the data and performing advanced analysis for optimization is difficult. Consequently, such advanced analysis tends to incur high economic, computational, and other costs.

[0004] Given these circumstances, there is a need to automate and support advanced analysis and other processing related to the optimization of human capital management, exemplified by forecasts of employee numbers reflecting regional conditions, on a distributed platform where information is securely managed.

[0005] Regarding technology for automating processes related to the optimization of human capital management, Patent Document 1 discloses a method for managing qualification information, characterized in that it receives search conditions for qualification-related information related to the facility management of a company to which a pre-registered user belongs via a network, extracts qualification-related information corresponding to the search conditions from at least one of the following: legal information and qualification information related to the facility management of the company, qualification information held by the company's employees, and information on the company's facility managers, and displays information on at least one of the following on the user's terminal screen: the age distribution of qualification holders in the company and a prediction of the temporal trend of the number of qualification holders.

[0006] The technology described in Patent Document 1 can display information regarding at least one of the following: the age distribution of qualified personnel in a company and a prediction of the temporal trend in the number of qualified personnel. [Prior art documents] [Patent Documents]

[0007] [Patent Document 1] Japanese Patent Publication No. 2003-162612 [Overview of the Initiative] [Problems that the invention aims to solve]

[0008] However, the technology described in Patent Document 1 is limited to performing analyses only on qualified personnel within a company. Furthermore, Patent Document 1 does not adequately describe the specific configuration and operation of the technology disclosed in the document in a way that would enable a person skilled in the art to implement it. Therefore, the document has room for further improvement in providing specific technical means to support and automate advanced analyses related to the optimization of human capital management on a distributed platform where information is securely managed.

[0009] The objective of this invention is to provide a technical means for automating and supporting advanced analysis related to the optimization of human capital management on a distributed platform where information is securely managed. [Means for solving the problem]

[0010] As a result of diligent research to solve the above problems, the inventors have found that the above objectives can be achieved by generating prediction results for the number of employees through probabilistic trials using a probabilistic model based on personnel information, and by reflecting future projections of population dynamics within the main commuting area of ​​employees in these probabilistic trials, as well as by other technical features. Thus, the inventors have completed the present invention.

[0011] One aspect of the present invention relates to the employees during the period in question. Includes age, place of residence, employment history, and employment history.A personnel information acquisition unit that acquires personnel information, and the personnel information The distribution of the residential locations of the employees included based on, Major commuting areas are identified as regions containing a predetermined percentage of employees, and population mesh data based on public statistics corresponding to those major commuting areas are used. A demographic data acquisition unit that acquires future projections of demographic trends, and the aforementioned personnel information Based on this, the number of hires and the number of resignations are treated as probabilistic events, and the first parameter of the probability distribution related to the number of hires and the second parameter of the probability distribution related to the number of resignations are included. Using a probability model multiple people A prediction unit that generates prediction results for the trend in the number of employees through probabilistic trials, The following is calculated by performing statistical processing on the multiple prediction results obtained from the multiple probabilistic trials: The system includes an output unit that outputs the average value and confidence interval for the aforementioned trend, The forecasting unit, based on the future projections of demographic trends, Using the corrected first parameter Perform the aforementioned probabilistic trial. This provides a support tool for analyzing human capital management.

[0012] In this embodiment, employee personnel information for the target period is acquired, and this personnel information is reflected in a probabilistic model used in probabilistic trials to predict the trend in the number of employees. This personnel information includes at least the number of new hires and the number of employees who have left the company, and the probabilistic model includes parameters related to these. As a result, this embodiment can make appropriate predictions about the trend in the number of employees by treating events related to the number of new hires and employees who have left the company as probabilistic events that are in line with the actual situation shown in the personnel information.

[0013] In addition, in this embodiment, future projections of population dynamics within major commuting areas are obtained based on personnel information in a probabilistic trial, and predictions are made using a first parameter corrected by these future projections. Then, in this embodiment, the mean value and confidence interval for trends are output, calculated by statistical processing of multiple prediction results, similar to the so-called bootstrap method or jackknife method. As a result, this embodiment can make predictions that take into account the uncertainty based on probabilistic trials.

[0014] Based on the above, this embodiment can provide a technical means to support the automation of advanced analysis related to the optimization of human capital management on a distributed platform where information is securely managed.

[0015] Furthermore, the present invention can take various aspects exemplified below: an aspect of improving prediction accuracy by using a so-called multi-agent model; an aspect of enhancing the granularity of prediction by tracking the temporal changes of agents having an age state; an aspect of performing advanced prediction combining age and attributes by tracking temporal changes including acquisition and loss of attributes in agents; an aspect of estimating processing time and calculating a consideration based on the processing time; an aspect configured to be able to efficiently generate an analysis result corresponding to the relevant data without leaking the user's mesh data to the outside by implementation using an API.

[0016] Due to the effects brought about by adding specific configurations to each of these aspects, these aspects contribute to providing technical means for automating and supporting advanced analysis related to optimization of human capital management on a distributed platform where information is securely managed.

Advantages of the Invention

[0017] As described above, the present invention can provide technical means for automating and supporting advanced analysis related to optimization of human capital management on a distributed platform where information is securely managed.

Brief Description of the Drawings

[0018] [Figure 1] FIG. 1 is a block diagram showing an example of the hardware configuration and software configuration of the system S of the present embodiment. [Figure 2] FIG. 2 is an example of employee mesh data 131. [Figure 3] FIG. 3 is an example of employment data 132. [Figure 4] FIG. 4 is an example of retirement data 133. [Figure 5] FIG. 5 is a data flow diagram schematically showing the data flow in the simulation of the present embodiment. [Figure 6] FIG. 6 is an explanatory diagram of the agent model in the simulation of the present embodiment. [Figure 7]Figure 7 is an explanatory diagram illustrating the schematic configuration of the simulation in this embodiment. [Figure 8] Figure 8 is a main flowchart showing an example of a preferred flow of support processing performed by the support device 1 of this embodiment. [Figure 9] Figure 9 is a continuation of the previous figure. [Modes for carrying out the invention]

[0019] Firstly, although the following disclosures, figures, and / or claims are described either individually or in combination with one or more other aspects, the subject matter of the immediate disclosure is not intended to be limited in that way. That is, the immediate disclosures, figures, and claims are intended to encompass the various aspects described herein, either individually or in one or more combinations with each other. For example, even if the immediate disclosure describes and illustrates the first, second, and third embodiments in such a way that the first embodiment is described and illustrated particularly in relation to the second embodiment, or the second embodiment is described and illustrated only in relation to the third embodiment, the immediate disclosures and illustrations are not limited in that way and may include only the first embodiment, only the second embodiment, only the third embodiment, or one or more combinations of the first, second, and / or third embodiments, such as the first and second embodiments, the first and third embodiments, the second and third embodiments, or the first, second, and third embodiments.

[0020] In this text, the phrase "or" is used to mean a "non-exclusive" arrangement unless explicitly specified otherwise. For example, when we say "item x is A or B," it means either (1) item x is either A or B, or (2) item x is both A and B. In other words, the word "or" is not used to define an "exclusive" arrangement.

[0021] Furthermore, when the phrases "contain at least one" or "contain at least one of the following" are used in the text, they mean that the system or element contains one or more of the elements listed after the phrase. For example, if there are three types of elements, from element 1 to element 3, the phrases "contain at least one" or "contain at least one of the following" are interpreted as any of the following structural arrangements: a device containing element 1, a device containing element 2, a device containing element 3, a device containing element 1 and element 2, a device containing element 1 and element 3, a device containing element 2 and element 3, or a device containing element 1, element 2, and element 3.

[0022] The same interpretation is intended when the phrase "used in at least one of the following" is used in the text. Furthermore, "and / or" as used in the text is used as a linguistic conjunction to indicate that one or more of the listed elements or conditions are included or occur. For example, a device containing the first element, the second element, and / or the third element is interpreted as any of the following structural arrangements: a device containing the first element, a device containing the second element, a device containing the third element, a device containing the first and second elements, a device containing the first and third elements, a device containing the second and third elements, or a device containing the first, second, and third elements.

[0023] Furthermore, the use of the phrase "and / or" in this text signifies a "non-exclusive" arrangement, as stipulated in the Japanese Industrial Standard (JIS) "Format and Preparation Method of Standards Documents JIS Z 8301".

[0024] The following describes in detail an example of an embodiment of the present invention with reference to the drawings.

[0025] <System S> Figure 1 is a block diagram showing an example of the hardware and software configuration of the support system (System S) of this embodiment. Hereinafter, an example of a preferred embodiment of System S of this embodiment will be described using Figure 1.

[0026] System S comprises a support device 1 for analysis related to human capital management. Preferably, the support device 1 is configured to communicate with a terminal T via a network N. The support device 1 may also have terminal functionality.

[0027] [Support device 1] Support device 1 comprises various hardware components such as a control unit 11, a storage unit 13, and a communication unit 14.

[0028] [Configuration as a Mesh Server] In order to perform efficient calculations using map data in mesh data format, it is preferable that the support device 1 be a mesh server that manages mesh data. The mesh data may be, for example, a primary mesh section (approximately 80 km per side), a secondary mesh section (approximately 10 km per side), a tertiary mesh section (standard regional mesh, approximately 1 km per side), a quaternary mesh section (approximately 500 m per side), a quintic mesh section (approximately 250 m per side), a quaternary mesh section (approximately 125 m per side), an extended 100 m mesh, an extended 10 m mesh, or an extended 1 m mesh, or various other regional meshes or meshes usable on a global scale. The mesh data may also be hierarchical mesh data that combines regional meshes or meshes usable on a global scale with different granularities.

[0029] (Regarding configuration as a distributed mesh data server) In order to enable rapid calculations on large amounts of mesh data, it is preferable that the support device 1 be configured as a distributed mesh data server including multiple server devices configured to perform parallel processing on mesh data.

[0030] A distributed mesh data server preferably divides the calculations for a large number of meshes among its servers and processes them in parallel. Hash partitioning, data range partitioning, and other conventional computation partitioning methods can be applied to the calculations. For example, if each server corresponds to or is associated with a region, a partitioning method is possible where the calculations are divided by region and assigned to each server.

[0031] To prevent data leakage between users, it is preferable for distributed mesh data servers to manage data access through login processing and encryption. To prevent data leakage during inter-server communication, it is preferable for distributed mesh data servers to encrypt communications using symmetric-key cryptography or public-key cryptography.

[0032] [Control Unit 11] The control unit 11 comprises a central processing unit (CPU), random access memory (RAM), read-only memory (ROM), and other hardware components.

[0033] The control unit 11 cooperates appropriately with at least one of the storage unit 13 and the communication unit 14 to realize the software components related to the program of this embodiment executed by the support device 1. The support device 1 includes a personnel information acquisition unit 111, a population dynamics acquisition unit 112, a prediction unit 113, and an output unit 116 as software components. The support device 1 may also include a processing time estimation unit 114, a compensation calculation unit 115, and other software components.

[0034] Details of the support processing implemented by the aforementioned software components will be explained later with reference to Figures 8 and 9.

[0035] [Storage Unit 13] The storage unit 13 is a device on which data and / or files are stored, and has a storage unit for non-temporarily storing data. The storage unit includes, for example, a hard disk, semiconductor memory, recording medium, and / or memory card or other storage material. The storage unit 13 stores a program that is executed by the control unit 11.

[0036] Furthermore, employee mesh data 131 is stored in the storage unit 13. In addition, it is preferable that recruitment data 132 and / or retirement data 133 are further stored in the storage unit 13.

[0037] (Employee Mesh Data 131) Employee mesh data 131 is statistical data that assigns employee data such as place of work, occupation, age, etc., to regional meshes, globally available meshes, or meshes that are hierarchical in nature (hereinafter sometimes abbreviated as "regional meshes, etc." or simply "mesh") based on place of residence. Employee mesh data 131 may further include other statistical data about employees (e.g., type of employment contract).

[0038] In order to realize a simulation that takes into account the workplace, job type, and place of residence, it is preferable that the employee mesh data 131 be classified by workplace and job type, and that it include statistical data showing the number of employees by age for each place of residence grouped by mesh.

[0039] Figure 2 shows an example of employee mesh data 131. This example visualizes statistical data showing the number of employees by age for each residential area grouped within each mesh, for employees working at workplace F, by overlaying it onto a map. Note that this example visualizes the correspondence between regional mesh data and corresponding statistical values ​​for illustrative purposes, and employee mesh data 131 is not limited to data that stores other data (e.g., map data) in addition to statistical data.

[0040] In the following examples, the mesh in row x, column Y may be abbreviated as "mesh xY" or simply "xY". For example, the mesh in row c, column B that includes work location F will be abbreviated as "mesh cB" or "cB".

[0041] In this example, statistical data on the age distribution is shown for six meshes, with mesh aA in the upper left corner and mesh bC in the lower right corner, as follows: Number of residents aged 18 to 25: A standard number (average of about 3 people per age group). Number of residents aged 62 to 65: A relatively large number (average of about 17 people per age group).

[0042] Furthermore, in this example, statistical data on the age distribution is shown for the four meshes, with mesh cB in the upper left corner and mesh dC in the lower right corner, as follows: Number of young residents: a relatively large number (average of about 5 people per age group). Number of elderly residents: a standard number (average of about 10 people per age group).

[0043] Furthermore, in this example, statistical data on the age distribution is shown for two meshes, mesh cA and mesh dA, as follows: Number of young residents: very small number (average of approximately 0.5 people per age group). Number of elderly residents: small number (average of approximately 3 people per age group).

[0044] By storing such statistical data in the employee mesh data 131, the support device 1 can identify the commuting area of ​​employees and perform human capital management analysis based on that commuting area. In the example, the support device 1 can identify six meshes with mesh aA as the upper left corner and mesh bC as the lower right corner, and four meshes with mesh cB as the upper left corner and mesh dC as the lower right corner, as commuting areas related to the workplace F, and perform analysis.

[0045] Furthermore, by storing statistical data as shown in the example, the support device 1 can perform analyses related to human capital management based on age distribution. In the example, the support device 1 can perform an analysis based on an age distribution of approximately 1:3 for young people to older people at work location F.

[0046] In addition, by storing statistical data as shown in the example, the support device 1 can perform analyses related to human capital management based on combinations of commuting areas and age distributions. In the example, for example, the support device 1 can perform an analysis in which, for workplace F, four meshes with mesh cB as the upper left corner and mesh dC as the lower right corner are defined as commuting areas for young people, and six meshes with mesh aA as the upper left corner and mesh bC as the lower right corner are defined as commuting areas for older people.

[0047] Furthermore, by storing statistical data that includes job categories omitted in the example, the support device 1 can perform analyses related to human capital management based on job categories. Similarly, the support device 1 can perform analyses based on job categories and commuting area, analyses based on job categories and age distribution, and analyses based on job categories, commuting area, and age distribution, respectively.

[0048] (Recruitment Data 132) Recruitment data 132 is statistical data on the number of recruits by year, work location, job type, and age. By storing such recruitment data 132, the support device 1 can perform analysis on human capital management based on recent recruitment trends. Furthermore, the support device 1 can perform analysis based on recruitment trends by job type. The format of recruitment data 132 is not particularly limited.

[0049] Figure 3 shows an example of recruitment data 132. This example shows recruitment results at work location F from 2020 to 2025. In this example, job categories are divided into "manual labor" and "technical / administrative labor." The data in this example records the total number of hires and the number of hires for each job category by year. For the sake of readability, the age of the hires has been omitted in this example.

[0050] In this particular case, the total number of hires was roughly just under 30 during the pandemic from 2020 to 2021. However, from 2022 onward, as the pandemic subsided, the total number of hires increased year by year, reaching 72 in 2024. Subsequently, in 2025, the total number of hires decreased to 60.

[0051] Looking at the data by job type, the number of employees hired for manual labor positions was 22 in 2020 and 21 in 2021, but has shown an increasing trend since 2022, reaching 60 in 2024. On the other hand, the number of employees hired for technical and administrative positions remained relatively stable throughout the entire period, staying within the range of 8 to 12.

[0052] By storing this recruitment data 132, the support device 1 can perform analyses of human capital management based on recruitment trends by job type and year. Furthermore, the support device 1 can reflect recruitment trends in simulations for manual laborers and simulations for technical and administrative workers.

[0053] (Retirement Data 133) Retirement data 133 is statistical data on the number of retirees by year, work location, job type, and age. By storing such retirement data 133, the support device 1 can perform analysis on human capital management based on retirement trends. Furthermore, the support device 1 can perform analysis based on retirement trends by job type. The format of retirement data 133 is not particularly limited.

[0054] Figure 4 shows an example of retirement data 133. This example shows retirement data for work location F from 2020 to 2025. In this example, job categories are divided into "manual labor" and "technical / clerical labor." The data in this example records the total number of retirees and the number of retirees by job category, year by year. For the sake of readability, the ages of retirees have been omitted in this example.

[0055] In this example, the total number of retirees was 172 in 2020, but gradually decreased from 2021 onwards, reaching 156 in 2024. This decrease in the number of retirees suggests that the mass retirement of the generation centered around those reaching age 65 has run its course and peak has passed. Furthermore, the number for 2025 is low at 61 because it is still the middle of the fiscal year.

[0056] Looking at the data by job type, the number of retirees in manual labor positions accounted for approximately 75 to 80% of all retirees each year, indicating that the majority of retirements are concentrated in manual labor positions. On the other hand, the number of retirees in technical and administrative positions remained stable at around 35 throughout the entire period, showing less year-to-year fluctuation compared to manual labor positions.

[0057] By storing this retirement data 133, the support device 1 can grasp retirement trends by year and by job type. Furthermore, the support device 1 can reflect these retirement trends in simulations for manual laborers and simulations for technical and clerical workers.

[0058] [Communication Unit 14] The communication unit 14 is not particularly limited as long as it connects the support device 1 to the network N and enables communication. Examples of the communication unit 14 include a network card compatible with the Ethernet standard and communication equipment compatible with wireless LAN.

[0059] [Regarding anonymization] The support device 1 of this embodiment may be configured to output the data processing results in an anonymized format. This prevents the leakage of personal information from the data processing results.

[0060] Let's explain using the example of preventing address leakage from mesh data. Support device 1, for example, performs a process to control the granularity of the mesh so that it satisfies a predetermined k-anonymity threshold k. If any mesh does not contain k addresses, this process involves, for example, merging that mesh into a coarser mesh so that all meshes contain k or more addresses.

[0061] As a result of this processing, a person viewing the data processing results can infer that one of the k or more addresses included in the mesh is the destination, but it becomes difficult to identify the specific address. l- To satisfy the diversity requirement, the support device 1 may perform similar processing to control the granularity of the mesh.

[0062] [Network N] The type of network N is not particularly limited, as long as it enables mutual communication between information processing devices included in system S. Examples of network N include the internet, a mobile phone network, and a wireless LAN.

[0063] [Terminal T] The type of terminal T is not particularly limited and includes, for example, a desktop personal computer, a laptop computer, a smartphone, a tablet device, and other terminal devices.

[0064] Terminal T is configured to perform, for example, the process of instructing the support device 1 to perform analysis, and the process of acquiring and displaying the analysis results from the support device 1. This process may be implemented by a program installed on terminal T, or by a browser that processes static and / or dynamic data provided by the support device 1.

[0065] [Simulation Outline] The following is an outline of the simulation that the support device 1 of this embodiment performs to support the analysis of human capital management.

[0066] [Data Flow] Figure 5 is a data flow diagram that schematically shows the data flow in the simulation of this embodiment. The following is a schematic explanation of the data flow in the simulation of this embodiment using Figure 5.

[0067] The top row of the figure shows employee data, recruitment data 132, and retirement data 133 as input data for the simulation.

[0068] As shown in the upper left of the figure, the support device 1 acquires employee data, converts this data into mesh data, and stores it as employee mesh data 131. Then, the support device 1 extracts mesh statistical data related to the simulation from the employee mesh data 131.

[0069] Furthermore, as shown in the upper center to the right of the figure, the support device 1 extracts recruitment data related to the simulation from the recruitment data 132. In addition, the support device 1 extracts retirement data related to the simulation from the retirement data 133.

[0070] As shown in the center right of the diagram, support device 1 acquires future projections of population dynamics within the main commuting area of ​​employees (e.g., future population projections from the Ministry of Land, Infrastructure, Transport and Tourism). Support device 1 then passes mesh statistical data, recruitment data, retirement data, and future projections as parameters to the agent model.

[0071] As shown below the figure, the support device 1 generates simulation result output data based on the simulation results output by the agent model and outputs it to a display or other output device.

[0072] [Agent Model] Figure 6 is an explanatory diagram of the agent model in the simulation of this embodiment. The following is a description of the agent model in the simulation of this embodiment using Figure 6.

[0073] The support device 1 of this embodiment predicts the trend in the number of employees through simulation using an agent model. In this agent model, agents correspond to employees or groups thereof and have at least one attribute: age. The simulation performs operations to change, add, or delete agents based on aging, increases in the number of employees (e.g., hiring, transfers), and decreases in the number of employees (e.g., resignations, transfers) that occur over time. The simulation then predicts the trend in the number of employees through these operations.

[0074] In this agent model, agents may be representative agents that group employees by the same or similar attributes, or they may be individual agents corresponding to individual employees. Representative agents further have an attribute of the number of people that changes with operations such as increasing or decreasing the number of people. Representative agents can also be generated from simulation results performed at business offices or departments. If it is desired to reduce the computational load in the simulation, it is preferable for the support device 1 to perform the simulation using representative agents.

[0075] Figure 6 shows an example of a representative agent, with attributes including age (18 to 65 years old) and number of employees. For clarity, agents corresponding to ages 26 to 61 years old are omitted from the figure.

[0076] In this example, each agent's attribute: number of employees increases due to events that increase the number of employees, such as hiring new graduates and mid-career hires. Conversely, each agent's attribute: number of employees decreases due to events that decrease the number of employees, such as changing jobs, death, and retirement. As time progresses in the simulation, support device 1 increases the agent's attribute: age in proportion to the passage of time.

[0077] In job-specific simulations, it is preferable that the agent also possesses attributes specific to that job. This allows the support device 1 to perform simulations that reflect different hiring and resignation trends for each job type.

[0078] In simulations for each work location, it is preferable that the agent further possesses attributes of the work location. This allows the support device 1 to realize simulations that reflect different hiring and resignation trends for each work location. In addition, the agent in this embodiment may further possess attributes such as department, employment contract type, and other attributes. This allows the support device 1 to perform simulations for employees classified by those attributes.

[0079] [Simulation Outline Configuration] Figure 7 is an explanatory diagram that schematically shows the configuration of the simulation in this embodiment. The following is a description of the schematic configuration of the simulation according to this embodiment, using Figure 7.

[0080] As shown in the upper left of the figure, the support device 1 converts the employee data into employee mesh data 131, which includes various meshes separated by attribute (e.g., age mesh, occupation mesh).

[0081] Furthermore, as shown in the upper right to center of the figure, the support device 1 performs aggregation processing on the recruitment data 132 and the resignation data 133 to generate a first parameter (new hire parameter) related to the number of new hires and a second parameter (resignation parameter) related to the number of resignations.

[0082] In the probabilistic model used in the simulation, the first parameter is generated, for example, based on the number of hires by age group over the past few years, which is stored in the hiring data 132. In the simulation, the support device 1 treats events of increased numbers due to hiring or other circumstances as accidental events following a Poisson distribution. Therefore, the support device 1 generates the first parameter as a parameter of the probabilistic model following a Poisson distribution corresponding to the number of hires mentioned above.

[0083] For example, the support device 1 has a first parameter λ i,j,k The equation λ i,j,k = 1 / 6Σ t=2019 2024 Injectable Pop i,j,k It is generated based on (t). However, i is a code indicating the business establishment, j is a code indicating the occupation, and k is a code indicating the age group. Also, injectionPop i,j,k (t) indicates the number of hires who meet the requirements related to i, j, and k.

[0084] In this example, the first parameter is generated by summing the number of successful hires from 2019 to 2024 and dividing by the number of years "6". In other words, this example generates the first parameter of the Poisson distribution for each age group related to a probabilistic model showing the temporal changes in a hypothetical employee with an age status.

[0085] In the probability model used in the simulation, the second parameter is generated based on, for example, the number of retirees for each age in the recent few years stored in the retirement data 133. In the simulation, the support device 1 treats the event of a decrease in the number of people due to retirement or other reasons as a random event following a Poisson distribution. Therefore, the support device 1 generates the second parameter as the parameter of the probability model following the Poisson distribution corresponding to the above-mentioned number of retirees.

[0086] For example, the support device 1 uses the second parameter ρ i,j,k and generates it based on the formula ρ i,j,k =1 / 6Σ t=2019 2024 exitPop i,j,k (t). Here, i is the code indicating the workplace, j is the code indicating the job type, and k is the code indicating the age group. Also, exitPop i,j,k (t) represents the number of retirees satisfying the requirements related to i, j, and k.

[0087] In this example, the second parameter is generated by accumulating the number of retirees satisfying the requirements from 2019 to 2024 and dividing by the number of years "6". That is, in this example, the second parameter of the Poisson distribution for each age group related to the probability model showing the change over time of virtual employees having the age status is generated.

[0088] "The recent few years" in the generation of the first parameter or the second parameter is not particularly limited. In order to balance ensuring a sufficient number of samples and not being dragged down by past data, "the recent few years" is preferably defined within the range of 3 to 10 years, more preferably within the range of 5 to 8 years, and most preferably 6 years. Also, when a simulation according to the job type or other attributes is performed, the first parameter and the second parameter are preferably generated for each such attribute.

[0089] By generating the first and second parameters as parameters of a probability model following a Poisson distribution, the support device 1 can reduce the computational complexity of the simulation through modeling based on the reasonable assumption that the probability of occurrence per unit of time is constant.

[0090] In the simulation, support device 1 corrects the first parameter based on future projections of population dynamics within the main commuting area of ​​employees (e.g., future population projections from the Ministry of Land, Infrastructure, Transport and Tourism). Then, using an agent simulator, support device 1 simulates the temporal changes of agents based on employee mesh data 131 through probabilistic trials using a probabilistic model with the corrected first and second parameters.

[0091] In other words, the support device 1 generates prediction results (simulation results) of the trend in the number of employees through probabilistic trials using a probabilistic model based on personnel information (recruitment data 132, retirement data 133).

[0092] At this time, the support device 1 executes the above simulation multiple times and generates multiple prediction results. Then, the support device 1 calculates the mean value and confidence interval for the trend through statistical processing of these multiple prediction results and outputs the results.

[0093] As a result, the support device 1 can output the mean value and confidence interval for the trend, calculated by statistical processing of multiple prediction results, similar to the so-called bootstrap method or jackknife method. Therefore, this embodiment can perform predictions that take into account uncertainty based on stochastic trials.

[0094] This embodiment reduces the computational complexity associated with the number of generation iterations by suppressing the computational complexity per simulation using a probability model based on a Poisson distribution. When the support device 1 is a distributed mesh server, it generates multiple prediction results through parallel processing. As a result, the support device 1 can perform predictions that take into account uncertainty based on probabilistic trials while keeping computation time down.

[0095] Personnel information is sensitive to privacy and must be managed securely. Support device 1, configured as a distributed mesh server, can perform information management such as access control and communication encryption. Therefore, such support device 1 can automate and support advanced analysis related to the optimization of human capital management on a distributed platform where information is securely managed.

[0096] [Main Flowchart of Support Processing] Figure 8 is a main flowchart showing an example of a preferred flow of support processing performed by the support device 1 of this embodiment. Figure 9 is a continuation of the previous figure. The following is an example of a preferred flow of support processing performed by the support device 1 of this embodiment, using Figures 8 and 9.

[0097] [Step S1: Determine whether to start the simulation] The control unit 11 works in cooperation with the memory unit 13 and the communication unit 14 to determine whether to start a simulation related to the analysis of human capital management (start determination step). If the control unit 11 determines to start, it moves the process to step S2; otherwise, it returns the process to step S1.

[0098] In this step, the control unit 11 determines to start the simulation, for example, when it receives a request to start the simulation from terminal T. In order to efficiently generate analysis results corresponding to the user's mesh data without leaking the user's mesh data to the outside, it is preferable that the support device 1 determines to start the simulation when it receives command data from terminal T via the API. At this time, it is preferable that the support device 1 is configured to acquire various data described later via the API. In order to prevent the user's mesh data from being leaked to the outside, it is preferable that the API encrypts the data it transmits.

[0099] [Step S2: Acquire personnel information] The control unit 11 works in cooperation with the memory unit 13 to execute the personnel information acquisition unit 111. The control unit 11 then uses the personnel information acquisition unit 111 to perform the process of acquiring employee personnel information for the target period (personnel information acquisition step). The control unit 11 then moves the process to step S3.

[0100] In this step, the personnel information acquisition unit 111 acquires, as the aforementioned personnel information, information including at least the number of employees by age, the number of new hires, and the number of retirees (e.g., employee mesh data 131, hiring data 132, and retirement data 133).

[0101] Furthermore, in this step, it is preferable that the personnel information acquisition unit 111 acquires information including, as described above, information relating to the employee's attributes, such as retention status, acquisition date, and loss date. This enables the support device 1 to perform simulations regarding the employee's attributes.

[0102] [Step S3: Obtain population dynamics] The control unit 11 works in cooperation with the memory unit 13 to execute the population dynamics acquisition unit 112. The control unit 11 then uses the population dynamics acquisition unit 112 to perform the process of obtaining future projections of population dynamics within the main commuting area of ​​the aforementioned employees (population dynamics acquisition step). The control unit 11 then moves the process to step S4.

[0103] In this step, the demographic data acquisition unit 112, for example, identifies an area containing a given percentage of employees who meet the requirements from the employee mesh data 131 as a major commuting area, and then performs the above-mentioned processing by a series of procedures to acquire future projection data corresponding to the major commuting area from mesh data showing future demographic projections (e.g., the Ministry of Land, Infrastructure, Transport and Tourism's future projection population mesh statistics).

[0104] The region described above is, for example, a circular region centered on the workplace that includes a given proportion of employees. The circular region is advantageous because it fits typical commuting patterns where employees' residences are concentrated in areas close to their workplaces, and the implementation of the process for determining the radius is easy, computationally intensive, and the representation is simple. In addition to the circular region, the region described above may also be a region represented by a set of meshes selected such that the residences of at least a given proportion of employees are included and the area is minimized. This region is advantageous because it is less likely to include extraneous areas when employees' residences are dispersed across multiple concentrated areas, and the region can be identified using a method with relatively low computational complexity (e.g., sorting the meshes and adding meshes to the region in descending order of the number of employees until the given proportion is exceeded).

[0105] The given percentage is not particularly limited as long as it corresponds to the identification of the main commuting area. From the viewpoint of balancing the area not being too large and far removed from actual commuting conditions with reducing the number of employees who fall outside the area, the given percentage is preferably 60% or more and less than 95%, more preferably 70% or more and less than 90%, even more preferably 75% or more and less than 85%, and the optimal value is 80%. The circular area at the optimal value is "D 80 It is also referred to as "within the circle."

[0106] The control unit 11 executes a process to construct a probabilistic model based on personnel information for generating prediction results of the trend in the number of employees, using the prediction unit 113 (described later). That is, the prediction unit 113 executes a process to construct a probabilistic model based on personnel information for generating prediction results of the trend in the number of employees. This probabilistic model includes at least a first parameter related to the number of new hires and a second parameter related to the number of employees who leave the company. The first parameter is corrected by future projections. Step S4 is an example of this process.

[0107] [Step S4: Constructing the Agent Model] The control unit 11 works in cooperation with the memory unit 13 to execute the prediction unit 113. Then, the control unit 11 executes the process of constructing the agent model using the prediction unit 113 (model construction step). The control unit 11 then moves the process to step S5.

[0108] In this step, the prediction unit 113 constructs an agent model by modeling the employee composition shown in the personnel information using virtual employees with age status (age attribute). Preferably, this agent model is a model using a Poisson distribution, as described in the "Outline Configuration of Simulation" section. This allows the support device 1 to reduce the computational complexity of the simulation through modeling based on the reasonable assumption that the probability of occurrence per unit of time is constant.

[0109] Furthermore, in order to reduce the computational load in the simulation, it is preferable that the agent model corresponds to the "representative agent" described in the "Agent Model" section.

[0110] If information including the retention status, acquisition timing, and loss timing of attributes is acquired as personnel information, it is preferable in this step for the prediction unit 113 to construct an agent model by modeling it with a virtual employee who further has the status related to the attributes. In this case, it is preferable that the probabilistic model related to the agent model is also a probabilistic model related to the acquisition and loss of attributes.

[0111] The support process preferably includes a process to estimate the processing time before the simulation starts and to calculate the expected compensation based on the processing time. This allows the user of the support device 1 to make a decision regarding the execution of the simulation based on the estimation result (e.g., whether to run the simulation as is or to run it with modified parameters). Steps S5 to S8 are an example of this process.

[0112] [Step S5: Estimating Processing Time] The control unit 11 works in cooperation with the storage unit 13 to execute the processing time estimation unit 114. The control unit 11 then uses the processing time estimation unit 114 to perform the process of estimating the processing time related to the prediction unit 113 described above (processing time estimation step). The control unit 11 then moves the process to step S6.

[0113] In this step, it is preferable that the processing time estimation unit 114 estimates the processing time related to the prediction unit 113 based on the history of said processing time. In this estimation, the processing time estimation unit 114 estimates the processing time using, for example, the following procedures.

[0114] (Procedure A) A procedure to estimate the processing time in the history with the closest parameters, including the number of times the prediction result was generated. (Procedure B) A procedure to obtain the estimation result by interpolation using processing times in multiple histories with similar parameters, including the number of times the prediction result was generated (e.g., linear interpolation, spline interpolation, interpolation using local weighted regression, polynomial regression interpolation, interpolation using Gaussian process regression, weighted average interpolation using k-nearest neighbors (k-NN)).

[0115] [Step S6: Calculate the expected compensation] The control unit 11 works in cooperation with the storage unit 13 to execute the compensation calculation unit 115. The control unit 11 then performs the process of calculating the compensation for generating the prediction result based on the processing time described above (predicted compensation calculation step). The control unit 11 then moves the process to step S7.

[0116] In this step, the consideration calculation unit 115 calculates the consideration by a process that includes, for example, a procedure of multiplying the estimated processing time by the system usage fee per processing time. As a result, the support device 1 can calculate a reasonable consideration that reflects the costs that depend on processing time, such as increased electricity charges resulting from the long-term occupancy of the system and the increased power consumption due to the increased processing time.

[0117] The price calculation unit 115 may calculate the price including not only the price that depends on the processing time, but also the price that does not depend on the processing time (e.g., a basic fee).

[0118] [Step S7: Present execution confirmation dialog] The control unit 11 performs a process in which the cost calculation unit 115 presents an execution confirmation dialog that asks the user whether or not to run the simulation in the situation in which the above-mentioned cost is expected (dialog presentation step). The control unit 11 then moves the process to step S8.

[0119] [Step S8: Determine whether to run the simulation] The control unit 11 uses the cost calculation unit 115 to perform a process to determine whether to run the simulation (execution determination step). If the control unit 11 determines to run the simulation, it moves the process to step S9; otherwise, it returns the process to step S1 and repeats the process from step S1 to step S11.

[0120] In this step, the cost calculation unit 115 determines whether to run the simulation, for example, by a procedure based on input indicating that the simulation should be run, such as via the execution confirmation dialog described above.

[0121] [Step S9: Execute Simulation] The control unit 11 executes the process of executing the simulation that was determined to be executed in the execution determination step by the prediction unit 113 (prediction step). The control unit 11 then moves the process to step S10.

[0122] In this step, the prediction unit 113 generates prediction results for the trend in the number of employees through probabilistic trials using a probabilistic model based on the personnel information described above. Specifically, the prediction unit 113 uses a probabilistic model that includes at least a first parameter related to the number of new hires and a second parameter related to the number of employees who leave, and generates prediction results using the first parameter corrected by future projections.

[0123] The prediction unit 113 preferably generates prediction results for the trend in the number of employees by probabilistic trials using multiple virtual employees corresponding to the number of employees. In other words, the prediction unit 113 preferably generates prediction results using an agent model (multi-agent model).

[0124] At this time, the prediction unit 113 predicts the number of adopters in the year t of the target year injectionPop i,j,k (t) is, for example, the first parameter λ i,j,k Poisson distribution along the line Pois(λ i,j,k The number of agents is changed over time according to the formula. In addition, the prediction unit 113 predicts the number of employees who leave the company in the year t that is the target of the prediction, exitPop i,j,k (t) is, for example, the second parameter ρ i,j,k Poisson distribution along the line Pois(ρ i,j,k The number of agents is changed over time, following the principle that the probability of occurrence per unit of time is constant. This allows the support device 1 to reduce the computational complexity of the simulation through modeling based on the reasonable assumption that the probability of occurrence per unit of time is constant.

[0125] The number of trials (ensemble size) in the probabilistic trial is not particularly limited. The lower limit, upper limit, and optimal value of the number of trials may be appropriately selected depending on the parameters related to the prediction. The following is an example of the lower limit, upper limit, and optimal value of the number of trials. In order to suppress the enormous amount of computation and achieve both prediction accuracy and the lower limit of the number of trials, it is preferable that the lower limit of the number of trials be 50 or more, more preferably 80 or more, and even more preferably 90 or more. Also, in order to suppress the enormous amount of computation and achieve both prediction accuracy and the upper limit of the number of trials, it is preferable that the upper limit of the number of trials be 200 or less, more preferably 150 or less, and even more preferably 120 or less. The optimal value of the number of trials is, for example, 100.

[0126] The agent model allows support device 1 to distinguish employees by various attributes (e.g., age, job type, work location, employment type) and to individually set behaviors (e.g., turnover rate, hiring rate) according to each attribute. Therefore, support device 1 can make predictions that are more in line with reality than a model based on a simple overall average.

[0127] In human capital management, analysis using macro models makes it difficult to visualize local and attribute-specific movements. With an agent model, support device 1 can reflect external factors such as regional population trend forecasts, in addition to the non-uniform state of employees (e.g., regional distribution, occupational bias, aging) and / or structural changes (e.g., accelerated retirement in some occupations), in the behavior of agents, thereby visualizing local and attribute-specific movements.

[0128] Furthermore, using an agent model, support device 1 is designed to probabilistically generate various events (e.g., resignation, hiring) for individual agents, enabling granular event simulations. This allows support device 1 to perform simulations that reflect, for example, annual changes in population composition.

[0129] In addition, the agent model allows support device 1 to easily track changes in state over time and perform simulations that are well-suited for scenario analysis. This enables support device 1 to perform sensitivity analysis to parameter changes (e.g., "when the turnover rate increases," "when hiring is suppressed," "when there is a relocation of work locations"), policy evaluation, and visualization of prediction differences.

[0130] Furthermore, when configured as a distributed mesh data server, the agent model allows the support device 1 to achieve highly parallel simulations by assigning processing nodes on an agent-by-agent basis. In this embodiment, each parallelized simulation is a simulation using a probability model based on a Poisson distribution. Therefore, the support device 1 can achieve both high simulation accuracy and suppression of increased computational load at each processing node. As a result, the support device 1 can appropriately automate and support advanced analysis related to the optimization of human capital management on a distributed platform where information is securely managed.

[0131] (Regarding the simulation of attributes) In this step, it is preferable that the prediction unit 113 predicts the changes over time of virtual employees who further have attribute-related states by multiple probabilistic trials using a probability model for the acquisition and loss of attributes, thereby generating prediction results for the changes in the number of employees corresponding to each attribute. This allows the support device 1 to perform analysis including tracking of attributes (e.g., skill acquisition, qualification acquisition, changes in household composition).

[0132] [Step S10: Calculate the fee based on the actual processing time] The control unit 11 executes a process to calculate the compensation based on the actual processing time using the compensation calculation unit 115 (actual compensation calculation step). The control unit 11 then moves the process to step S11. The "actual processing time" referred to here corresponds to, for example, CPU time, which is defined as the sum of user time and system time.

[0133] Furthermore, when the support device 1 is configured as a distributed mesh data server, it is preferable that the "actual processing time" is the sum of the actual processing times at each processing node. This allows the support device 1 to reflect the total processing time across all processing nodes in its pricing. The "sum of actual processing time at each processing node" is, for example, the sum of the CPU time at each node.

[0134] In this step, the consideration calculation unit 115 calculates the consideration by a process that includes, for example, a procedure of multiplying the estimated processing time by the system usage fee per processing time. This allows the support device 1 to calculate a reasonable consideration that reflects the cost factors in resource-based billing, such as increased electricity charges resulting from the long-term occupancy of the system and increased power consumption due to increased processing time.

[0135] Furthermore, the fee calculation unit 115 may calculate the fee including not only fees that depend on processing time, but also fees that do not depend on processing time (e.g., basic charges, storage usage fees, communication bandwidth usage fees, external data acquisition costs, maintenance and management fees, security management fees). This allows the support device 1 to reflect other cost factors (e.g., cost factors other than processing time in resource-based billing, cost factors related to administrative procedures) in the fee. In addition, in order to calculate fees according to various contract types, it is preferable for the fee calculation unit 115 to reflect the fee structure for each contract type in the fee calculation.

[0136] [Step S11: Output Simulation Results] The control unit 11 works in cooperation with the storage unit 13 and the communication unit 14 to execute the output unit 116. The control unit 11 then uses the output unit 116 to execute the process of outputting the above-mentioned simulation results (output step). The control unit 11 returns the process to step S1 and repeats the process from step S1 to step S11.

[0137] In this step, it is preferable that the output unit 116 outputs the mean value and confidence interval for the trend in the number of employees, which are calculated by statistical processing of the multiple prediction results obtained in the procedure described above. This allows the support device 1 to perform predictions with uncertainty assessment based on probabilistic trials, similar to the so-called bootstrap method or jackknife method.

[0138] [Effects of the support process] In the support process described above, personnel information of employees during the target period is acquired (step S2), and this personnel information is reflected in a probabilistic model used for probabilistic trials to predict the trend in the number of employees (step S4). This personnel information includes at least the number of new hires and the number of employees who have left the company, and the probabilistic model includes parameters related to these (steps S2 and S4). As a result, the support device 1 that performs the support process described above can make an appropriate prediction of the trend in the number of employees by using probabilistic trials that treat events related to the number of new hires and employees who have left the company as probabilistic events that are in line with the actual situation shown in the personnel information (step S9).

[0139] In addition, the support process described above obtains future projections of population dynamics within the main commuting area based on personnel information in a probabilistic trial (step S3), and makes predictions using the first parameter corrected by these future projections (steps S4, S9). Then, the support process outputs the mean value and confidence interval for the trend calculated by statistical processing of multiple prediction results, similar to the bootstrap method or jackknife method (step S11). This allows the method to make predictions that take into account the uncertainty based on probabilistic trials.

[0140] Based on the above, the aforementioned support processing and the support device 1 that executes it can provide a technical means to automate and support advanced analysis related to the optimization of human capital management on a distributed platform where information is securely managed.

[0141] In addition, the above-mentioned support processing can take the form of a so-called multi-agent model (steps S4 and S9). This allows the support device 1, which executes the above-mentioned support processing, to distinguish employees by various attributes (e.g., age, job type, work location, employment type) and to individually set behaviors (e.g., turnover rate, hiring rate) according to each attribute. Therefore, the support device 1 can make predictions that are more in line with reality than a model based on a simple overall average.

[0142] In human capital management, analysis using macro models makes it difficult to visualize local and attribute-specific movements. The support device 1 in this embodiment reflects external factors such as regional population trend forecasts, in addition to the non-uniform state of employees (e.g., regional distribution, occupational bias, aging) and / or structural changes (e.g., accelerated retirement in some occupations), in the behavior of agents, thereby enabling the visualization of local and attribute-specific movements.

[0143] Furthermore, when configured as a distributed mesh data server, the support device 1, which uses a multi-agent model, can achieve highly parallel simulations by assigning processing nodes on an agent-by-agent basis. In this embodiment, each parallelized simulation is a simulation using a probability model based on a Poisson distribution. Therefore, the support device 1 can achieve both high simulation accuracy and suppression of increased computational load at each processing node. As a result, the support device 1 can appropriately automate and support advanced analysis related to the optimization of human capital management on a distributed platform where information is securely managed.

[0144] In this embodiment, the above-described support processing can be configured to improve the granularity of predictions by tracking changes over time in agents having an age status (steps S4 and S9). As a result, the support device 1 in this embodiment is designed to probabilistically cause various events (e.g., retirement, hiring) to occur for individual agents, enabling fine-grained event simulations. Therefore, the support device 1 can, for example, perform simulations that reflect annual changes in population composition.

[0145] In addition, the support device 1 in this embodiment facilitates tracking of state changes over time and enables simulations that are easily compatible with scenario analysis. As a result, the support device 1 can perform sensitivity analysis to parameter changes (e.g., "when the turnover rate increases," "when hiring is suppressed," "when there is a reassignment of work locations"), policy evaluation, and visualization of prediction differences.

[0146] Furthermore, the above-described support processing may include tracking changes over time, including the acquisition and loss of attributes in the agent (steps S4 and S9). This enables the support device 1, which performs the above-described support processing, to make advanced predictions combining age and attributes.

[0147] Furthermore, the support processing described above may take the form of estimating processing time and calculating compensation based on processing time (steps S5 to S6, step S10). This allows users of the support device 1 performing this form of support processing to make decisions regarding simulation execution based on the estimation results (e.g., whether to run the simulation as is or to run it with modified parameters). In addition, the support device 1 performing this form of support processing contributes to calculating and billing a reasonable compensation that reflects the cost factors in resource-based billing.

[0148] Based on the above, the aforementioned support processing and the support device 1 that executes it can provide a technical means to automate and support advanced analysis related to the optimization of human capital management on a distributed platform where information is securely managed.

[0149] <Example of Use> The following is an example of using System S of this embodiment.

[0150] [Uploading various data] The user uploads data in CSV format (employee data including the number of employees by age, recruitment data 132, and retirement data 133) to the support device 1 via terminal T. The support device 1 retrieves the uploaded data and stores it in the storage unit 13. The support device 1 further retrieves future population projection data for the target area.

[0151] [Instructions for Pre-processing] The user instructs the support device 1 via terminal T to perform pre-processing to predict the cost of a simulation used for human capital management analysis. The support device 1 constructs an agent model related to the simulation and predicts the processing time based on the model and the number of trials. Then, the support device 1 calculates the predicted cost and displays it on terminal T.

[0152] [Instructions for executing this process] The user confirms the displayed predicted value of the consideration and instructs the system to execute the simulation process. Support device 1 performs multiple simulations using the agent model and displays the resulting statistics on terminal T. The user uses these statistics to understand the demographic trends of employees in the company and utilizes them for analysis related to human capital management.

[0153] Within the scope of the concept of the present invention, those skilled in the art can conceive of various modifications and alterations. Therefore, such modifications and alterations are understood to fall within the scope of the present invention. For example, any addition, deletion, or design change of components, or addition, omission, or modification of processes, made by a person skilled in the art to the above-described embodiments, is also included within the scope of the present invention, as long as it retains the gist of the present invention. [Explanation of symbols]

[0154] S System 1 Support equipment 11 Control Unit 111 Human Resources Information Acquisition Department 112 Demographics acquisition department 113 Prediction Section 114 Processing time estimation unit 115. Price Calculation Section 116 Output section 13 Storage section 131 Employee Mesh Data 132 Recruitment Data 133 Retirement Data 14 Communications Department N Network T terminal

Claims

1. The Human Resources Information Acquisition Unit acquires personnel information including the age, place of residence, hiring history, and resignation history of employees during the target period. A population dynamics acquisition unit identifies a major commuting area as an area containing a predetermined percentage of employees based on the distribution of employee residences included in the aforementioned personnel information, and obtains future population dynamics projections from population mesh data based on public statistics corresponding to the major commuting area. Based on the aforementioned personnel information, the prediction unit treats the number of new hires and the number of employees leaving as probabilistic events, and generates a prediction result of the trend in the number of employees through multiple probabilistic trials using a probabilistic model that includes a first parameter of the probability distribution related to the number of new hires and a second parameter of the probability distribution related to the number of employees leaving. An output unit that outputs the mean value and confidence interval for the trend calculated by performing statistical processing on the multiple prediction results obtained from the multiple probabilistic trials, Equipped with, The prediction unit performs the probabilistic trial using the first parameter corrected based on the future projection of population dynamics. A support tool for analyzing human capital management.

2. The prediction unit configures virtual employees as agents, each having the age and employment status of the employees as attributes, and generates a number of such virtual employees corresponding to the number of employees. By performing the probabilistic trial, it generates a prediction result of the trend in the number of employees. The support device according to claim 1.

3. The prediction unit divides the age status of the virtual employees into age groups, and uses a probability model based on a Poisson distribution set for each age group to predict the changes in the state of the virtual employees over time through the multiple probabilistic trials, thereby generating the prediction result. The support device according to claim 2.

4. The aforementioned personnel information acquisition unit acquires information as personnel information, including the status of retaining attributes such as job title, work location, and employment type related to the employee, as well as the time of acquisition and loss of such attributes. The prediction unit generates prediction results for the change in the number of employees corresponding to each attribute by performing multiple probabilistic trials using a probabilistic model that treats the acquisition and loss of attributes as probabilistic events for the virtual employees having the attributes as state variables. The support device according to claim 3.

5. A processing time estimation unit estimates the processing time required for the execution of the probabilistic trial by the prediction unit based on the history of said processing time, A cost calculation unit calculates the cost for generating the prediction result based on the estimated processing time, The support device according to claim 1, further comprising:

Images