Talent team structure prediction method based on Monte Carlo method

Abstract

The invention provides a talent team structure prediction method based on a Monte Carlo method, and the method comprises the steps: building a probability density function of personnel data changing with time through the personnel data of an employer in a certain time period, and carrying out the random testing on a computer through the Monte Carlo method, and simulating a random process of personnel data change in a period of time, and further predicting the talent team structure change of the employer, thereby helping the employer to formulate a talent team construction plan. According to the method, various factors influencing the talent team structure change are comprehensively considered, and the scientificity and accuracy of talent team structure prediction are improved.

Description

[0001] Technology Neighborhood

[0002] The invention relates to the field of human resource planning technology, and in particular to a talent team structure prediction method based on the Monte Carlo method. Background Art

[0003] Human resource planning is to formulate a human resource supply and demand balance plan for the employer based on the changes in the internal and external environment and conditions of the employer, through the analysis and estimation of the employer's future human resource needs and supply conditions, so as to achieve the rational allocation of human resources and other resources, and effectively motivate and develop employees. Party and government agencies, state-owned enterprises and institutions usually formulate talent development plans to propose the main goals, specific tasks and guarantee measures for talent team building in the future period. The formulation of talent team planning requires the prediction of the future talent team structure, the comparison and analysis of talent demand and supply, and the determination of the possible talent shortage or surplus problems in specific fields or positions of the employer, so as to provide a basis for subsequent personnel management decisions.

[0004] The current talent team structure prediction methods mostly use the Markov chain model, which is relatively rough in the calculation of transition probability. Generally, the average transition probability is calculated for a category of positions as a whole, and relatively few influencing factors are considered. The calculation of the transition probability is not precise enough, resulting in insufficient accuracy in solving practical problems, especially for employers with large scale, complex job settings, and many types of personnel. It is difficult to achieve the accuracy required to solve practical problems.

[0005] The inventors of this application have discovered through research in the process of realizing the present invention that the Monte Carlo method is a numerical calculation method based on random sampling, which is used to estimate or solve complex problems. It is based on the ideas of random sampling and statistical inference, and obtains numerical approximate solutions by simulating a large number of random experiments or random sampling. Through the Monte Carlo method, simulation experiments are carried out to simulate the state changes of each staff member, and then the changes in the overall talent team structure of the employer in a certain period of time in the future are estimated. First, the personnel data of the employer in a certain period of time, including age, length of service, gender, political status, position, rank, title, etc., are obtained, and the probability density function of the personnel data changing over time is established. Then, the Monte Carlo method is used to conduct random experiments on the computer to simulate the random process of personnel data changes over a period of time, and then the talent team structure of the employer is predicted, thereby helping the employer to formulate a talent team construction plan. Summary of the invention

[0006] The purpose of the present invention is to provide a talent team structure prediction method based on the Monte Carlo method, which regards each personnel data of each staff member of the employer as a random variable, and the change of personnel data over time as a random process, establishes a probability density function that changes over time, and uses the Monte Carlo method to simulate the random process of personnel data changes on a computer, which can calculate more influencing factors and simulate the changes of personnel data more finely.

[0007] A method for predicting talent team structure based on Monte Carlo method, comprising the following steps:

[0008] Obtain the personnel data of the employer in a certain period of time and establish the talent resource space H and constraint conditions C;

[0009] Define the state space X τ ;

[0010] Define the action space S, S = {S a} a∈A , S a are all actions in state a;

[0011] Through the talent resource space H and the constraint condition C, the probability density function P is established, P = {p(s)}, where p(s) represents the probability of state action s(a'|a), s(a'|a) represents the action corresponding to state a entering state a' is s, where a∈A, a'∈A, s∈S a , A is the set of all states;

[0012] Use the Monte Carlo method to conduct random experiments on computers: Generate random numbers on the computer and combine the probability density p(s) of each state change to simulate the state changes of the workers;

[0013] According to the Monte Carlo simulation results, statistical analysis Changes in staff structure, X 0 is the state set of the staff at the initial moment, where T is the simulation time interval, X T is the set of worker status after T time interval.

[0014] Furthermore, the talent resource space H is a collection of information on the current status and historical status of a unit's staff, including age, length of service, gender, political status, position, rank, title, etc.; the constraints C are national policies, laws and regulations, and unit rules and regulations that affect changes in the structure of the talent team, including the statutory retirement age, the number of establishments of the employer, the number of leadership positions, staff introduction rules, and promotion rules.

[0015] Furthermore, X τIt is a subset of the talent resource space H, including the employment years (e), age (y), status x of all staff at time τ ij ;x ij ∈X, is the jth state of the i-th staff member; is the j-th state of the i-th staff member at time τ.

[0016] Furthermore, S a It is a set of random events, including promotion of one or more levels, promotion of one or more levels of professional title, demotion of one or more levels, recruitment, transfer, resignation, and retirement.

[0017] Furthermore, the establishment of the probability density function p(s) specifically includes the following two situations:

[0018] (1) Establish the probability density function of uncertain state transition: Take the staff status change years as the dependent variable and all staff attributes in the human resource space H as the grouping variables to conduct a one-way variance analysis to find out the staff attributes that are statistically significant for the staff status change. Divide the state change s into corresponding subsets according to the different levels of different staff attributes, and classify and calculate the probability density function of the state change s p(s) = f ij (t), the sample size for establishing the probability density function is greater than 30. If the sample size of some groups is insufficient, the groups are merged;

[0019] Use kernel density estimation to construct the jth group probability density function of the i-th state change

[0020]

[0021] in:

[0022] t k is the sample in this group;

[0023] n is the sample size of the classification subset;

[0024] h is a fixed bandwidth parameter, calculated using Scott's law, as follows:

[0025] σ is the standard deviation of the data, L is the number of samples in the subset,

[0026] K(x) is the Gaussian kernel function:

[0027]

[0028] (2) Establish the probability density function of deterministic state change p(s) = f i (x):

[0029]

[0030] Furthermore, the random experiment is performed on a computer using the Monte Carlo method: random numbers are generated on the computer, and combined with the probability density p(s) of each state change, the state change of the staff is simulated, specifically including:

[0031] (1) Generate a random number, for each state x ij Generate a random number r∈[0,α] through a random number generator, where α is the maximum proportion of the state change sample.

[0032] (2) Generate a new state according to the following rules;

[0033] (2.1) When an uncertain state change is not restricted by constraints, a corresponding state change s(a'|a) occurs, and the state x of the worker is changed to ij Adjust to a';

[0034] (2.2) For the uncertain state change s(a'|a) subject to constraints, the state of the m workers with the largest rp(s(a'|a)) in the same state is adjusted to a', where m is the maximum adjustment amount subject to constraints;

[0035] (2.3) For a deterministic state change s(a'|a), if the constraint is satisfied, the state of the worker is adjusted to a', otherwise it remains unchanged;

[0036] (2.4) State changes associated with time

[0037] e'=e+1

[0038] y'=y+1

[0039] Through the above process, a random process of 1 unit time is completed

[0040] (3) Repeat (1) and (2) T times, = complete the random process;

[0041] Repeat steps (1), (2) and (3) to generate multiple sampling simulations until the required accuracy is achieved.

[0042] The present invention uses probability density function to estimate the changes in personnel data, which can solve the problem that the existing technology is not precise enough in calculation; at the same time, the Monte Carlo method is used to simulate the state change process of each staff member, which greatly improves the scientificity and accuracy of the talent team structure prediction and can effectively improve the quality of talent team construction planning. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1is a schematic diagram of the state space of a staff member according to an embodiment of the present invention;

[0044] Figure 2 It is a state change probability grouping calculation diagram of an embodiment of the present invention;

[0045] Figure 3 It is a flow chart of a method for predicting talent team structure based on the Monte Carlo method according to an embodiment of the present invention;

[0046] Figure 4 The current structure of the talent team;

[0047] Figure 5 is the result of one-way ANOVA;

[0048] Figure 6 (a) is the probability density distribution of the years of service from intermediate professional title to deputy senior professional title in an embodiment of the present invention (graduate student); (b) is the probability density distribution of the years of service from intermediate professional title to deputy senior professional title in an embodiment of the present invention (undergraduate); (c) is the probability density distribution of the years of service from section level to deputy division level in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the present invention, but the protection scope of the present invention is not limited to the factual examples.

[0050] See also Figure 3 The embodiment of the present invention proposes a talent team structure prediction method based on the Monte Carlo method, which calculates the probability of each state change of each staff member, uses the Monte Carlo method to simulate the change process of the staff member's state, and predicts the structural changes of the employer's talent team.

[0051] An embodiment is given below:

[0052] This example considers the structural changes in the talent pool of 144 staff members of a public institution in 2018 over a five-year period.

[0053] Step 1: Obtain personnel data such as the current status and historical status of the unit’s staff, and establish the talent resource space H and constraint conditions C.

[0054] Step 2: Establish the current state space X of the staff 0 (See Figure 4 ), the status includes the staff's length of service (e), age (y), job title (x1), and position (x2). The status set A = {senior, deputy senior, intermediate, junior, director, deputy director, section, deputy section}

[0055] Step 3: Define the action space state: After analyzing the state space X of the staff member 0The action space S = {s1 (deputy senior|intermediate), s2 (deputy division|section level)}, where S 中级 ={s(deputy senior|middle level)}, S section level ={s(deputy division level|section level)}

[0056] Step 4: Establish a probability density function that changes over time. Perform a one-way ANOVA on all information in H to find out the variables that are statistically significant with the state transition. s (Deputy Senior | Intermediate) is related to the educational status (p < 0.05) (see Figure 5 ).

[0057] according to Figure 2 Find the probability density function f for action s (associate senior | intermediate) by academic level 11 (t), f 12 (t), the results are shown in Figure 6 (a) and (b) in .

[0058] No statistically significant variables were found for the state transition s(deputy division level|section level). The probability density function f2(t) was calculated for all samples. The results are shown in Figure 6 (c) in .

[0059] Step 5: Use the Monte Carlo method to simulate the status changes of staff.

[0060] (1) Generate a random number r for each state of each staff member ij ∈[0,α]. For undergraduates, α=0.206 for s(deputy senior|intermediate), α=0.192 for postgraduates, and α=0.11 for s(deputy division|section level).

[0061] (2) Action s (Deputy Senior | Intermediate) Postgraduate staff use function f 11 (e) The probability of being promoted to deputy senior ij , use function f for staff with bachelor's degree 12 (e) The probability of being promoted to deputy senior ij When r ij <p ij The status is adjusted from intermediate to associate senior (see Figure 1 ).

[0062] (3)s(Deputy Division Level|Section Level) is subject to the number of cadre positions. When the number of positions is vacant, max(f i2 (e)-r i2 )The corresponding staff member i status is adjusted from section level to deputy division level.

[0063] (4) The status of the corresponding staff member i is adjusted to retired.

[0064] (5)e 1 =e 0 +1,y 1 =y 0 +1, complete a random process of 1 unit time

[0065] (6) Repeat steps (1)-(5) 5 times to complete the random process

[0066] (7) Repeat steps (1)-(6) 27 times, and s (deputy senior level | intermediate level) meets the preset requirements. Repeat steps (1)-(6) 50 times, and s (deputy division level | section level) meets the preset requirements.

[0067] Step 6: Statistical analysis. According to the Monte Carlo simulation results after 5 years, 3 people will retire from the unit; there is an 83.3% chance that 42-43 staff members will be promoted from intermediate professional titles to deputy senior professional titles; 3 people will be promoted from section level to deputy division level, with an average age of 45.4 years. The simulation results are basically consistent with the talent structure of the institution in 2023.

[0068] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by technicians in the technical field within the technical scope disclosed in the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for predicting talent team structure based on the Monte Carlo method, characterized in that: The steps include: Obtain the personnel data of the employer in a certain period of time and establish the talent resource space H and constraint conditions C; Define the state space X τ ; Define the action space S, S = {S a } a∈A , S a are all actions in state a; Through the talent resource space H and the constraint condition C, the probability density function P is established, P = {p(s)}, where p(s) represents the probability of state action s(a'|a), s(a'|a) represents the action corresponding to state a entering state a' is s, where a∈A, a'∈A, s∈S a , A is the set of all states; Use the Monte Carlo method to conduct random experiments on computers: Generate random numbers on the computer and combine the probability density p(s) of each state change to simulate the state changes of the workers; According to the Monte Carlo simulation results, statistical analysis Changes in staff structure, X 0 is the state set of the staff at the initial moment, where T is the simulation time interval, X T is the set of worker status after T time interval.

2. The method for predicting the structure of a talent team based on the Monte Carlo method according to claim 1, characterized in that: The talent resource space H is a collection of information on the current status and historical status of a unit's staff, including age, length of service, gender, political status, position, rank, title, etc.; the constraints C are national policies, laws and regulations, and unit rules and regulations that affect changes in the structure of the talent team, including the statutory retirement age, the number of establishments of the employer, the number of leadership positions, staff introduction rules, and promotion rules.

3. The method for predicting the structure of a talent team based on the Monte Carlo method according to claim 1, characterized in that: X τ It is a subset of the talent resource space H, including the employment years (e), age (y), status x of all staff at time τ ij ;x ij ∈X, is the jth state of the i-th staff member; is the j-th state of the i-th staff member at time τ.

4. The method for predicting the structure of a talent team based on the Monte Carlo method according to claim 1, characterized in that: S a It is a set of random events, including promotion of one or more levels, promotion of one or more levels of professional title, demotion of one or more levels, recruitment, transfer, resignation, and retirement.

5. The method for predicting the structure of a talent team based on the Monte Carlo method according to claim 1, characterized in that: The establishment of the probability density function p(s) specifically includes the following two situations: (1) Establish the probability density function of uncertain state transition: Take the staff status change years as the dependent variable and all staff attributes in the human resource space H as the grouping variables to conduct a one-way variance analysis to find out the staff attributes that are statistically significant for the staff status change. Divide the state change s into corresponding subsets according to the different levels of different staff attributes, and classify and calculate the probability density function of the state change s p(s) = f ij (t), the sample size for establishing the probability density function is greater than 30. If the sample size of some groups is insufficient, the groups are merged; Use kernel density estimation to construct the jth group probability density function of the i-th state change in: t k is the sample in this group; n is the sample size of the classification subset; h is a fixed bandwidth parameter, calculated using Scott's law, as follows: σ is the standard deviation of the data, L is the number of samples in the subset, K(x) is the Gaussian kernel function: (2) Establish the probability density function of deterministic state change p(s) = f i (x):

6. The method for predicting the structure of a talent team based on the Monte Carlo method according to claim 1, characterized in that: The random experiment is conducted on a computer using the Monte Carlo method: random numbers are generated on the computer, and combined with the probability density p(s) of each state change, the state change of the staff is simulated, specifically including: (1) Generate a random number, for each state x ij Generate a random number r∈[0,α] through a random number generator, where α is the maximum proportion of the state change sample. (2) Generate a new state according to the following rules; (2.1) When an uncertain state change is not restricted by constraints, a corresponding state change s(a'|a) occurs, and the state x of the worker is changed to ij Adjust to a'; (2.2) For the uncertain state change s(a'|a) subject to constraints, the state of the m workers with the largest rp(s(a'|a)) in the same state is adjusted to a', where m is the maximum adjustment amount subject to constraints; (2.3) For a deterministic state change s(a'|a), if the constraint is satisfied, the state of the worker is adjusted to a', otherwise it remains unchanged; (2.4) State changes associated with time e'=e+1 y'=y+1 Through the above process, a random process of 1 unit time is completed (3) Repeat (1) and (2) T times, = complete the random process; Repeat steps (1), (2) and (3) to generate multiple sampling simulations until the required accuracy is achieved.

Images