CN110659757A - Employee departure prediction method and device - Google Patents
Employee departure prediction method and device Download PDFInfo
- Publication number
- CN110659757A CN110659757A CN201810696834.9A CN201810696834A CN110659757A CN 110659757 A CN110659757 A CN 110659757A CN 201810696834 A CN201810696834 A CN 201810696834A CN 110659757 A CN110659757 A CN 110659757A
- Authority
- CN
- China
- Prior art keywords
- sample
- departure
- index
- employee
- tendency
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/10—Office automation; Time management
- G06Q10/105—Human resources
- G06Q10/1053—Employment or hiring
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- Game Theory and Decision Science (AREA)
- Data Mining & Analysis (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a staff leave prediction method and a device, and relates to the technical field of deep learning, wherein the method comprises the following steps: the method comprises the steps of generating a training sample based on the leaving index data, the saving factors and the leaving tendency of staff samples, training a preset deep learning model to obtain a leaving prediction model, and inputting the leaving index data of a predicted staff into the leaving prediction model to obtain a leaving tendency value and saving factor strength of the predicted staff. The employee job leaving prediction method and the employee job leaving prediction device can predict and obtain employees with strong job leaving tendency, can effectively support human resource departments to screen and implement targeted saving measures by combining with human resource reservation strategies, further promotes the significant reduction of the enterprise employee job leaving rate, is used for supporting human loss and human resource reservation measures, can save the human cost of enterprises, and ensures the normal operation or work progress of the enterprises.
Description
Technical Field
The disclosure relates to the technical field of deep learning, in particular to a staff position-leaving prediction method and device.
Background
With the enthusiasm of the domestic internet and the electronic commerce people for environmental competition, the competition of the industry for human resources is strengthened. The time and energy spent by the enterprise on the recruiters continuously greatly increase the cost of human resources. Due to the reasons of leaving work, relieving employment and the like, the post is vacant, and therefore the enterprise can adjust or recruit the post and generate corresponding cost for the replacement of the personnel, which is considerable. How to reduce the rate of leaving time more reasonably becomes an important subject of a human resource system. By constructing a departure warning and talent reservation model and the like, a reservation strategy suggestion can be found and given in advance before employees possibly generate a departure idea, and key talents are reserved and excellent employee loss is reduced by targeted reservation measures such as deep communication, development plan formulation, care and the like.
The staff departure warning system generally has the following characteristics: the problem is complicated: large and medium-sized enterprises usually use the staff leave warning system consciously, and the staff have more staff, more departments, more posts, complex individual characteristics of the staff and the like; decision support: the post-leaving early warning system not only needs to inform the enterprise of possible post-leaving risks of personnel, but also needs to give main factors and strength of post-leaving of the personnel so as to help the enterprise to implement a saving measure. Current departure warning systems have a number of disadvantages: the model has poor interpretability, low accuracy, difficult guarantee of objectivity, incapability of providing the influence degree of each factor on case job leaving risk, and difficulty in processing large-scale data. Therefore, a new technical solution for employee job departure prediction is needed.
Disclosure of Invention
In view of the above, an object of the present invention is to provide a method and an apparatus for predicting employee departure.
According to one aspect of the present disclosure, there is provided an employee departure prediction method, including: generating a training sample based on sample job leaving index data, effective saving factors and job leaving tendencies corresponding to the employee sample; training a preset deep learning model by using a deep learning method based on the training samples to obtain an out-of-position prediction model; updating the preset deep learning model into the departure prediction model, and inputting predicted departure index data of the predicted staff into the departure prediction model to obtain the departure tendency value and the saving factor strength of the predicted staff; and determining the departure tendency and the departure factor of the predicted employee according to the departure tendency value and the saving factor strength.
Optionally, the training a preset deep learning model based on the training samples includes: taking the sample job leaving index data as feature data, and taking the effective saving factor and the job leaving tendency corresponding to the sample job leaving index data as initial prediction results; and training the preset deep learning model based on the feature data and the initial prediction result.
Optionally, the preset deep learning model includes: a three-layer neuron model; the three-layer neuron model comprises: an input layer neuron model, a middle layer neuron model and an output layer neuron model; the output of each layer of neuron model is used as the input of the next layer of neuron model; wherein the neurons of the input layer neuron model correspond to the sample due indicators data, and the neurons of the output layer neuron model correspond to the retention factors and the tendency to cause a failure.
Optionally, the three-layer neuron model is a sub-network structure of a plurality of neural network layers having a fully connected structure; wherein the middle layer neuron model is a full-link layer.
Optionally, the employee sample comprises: the training samples are generated based on sample departure index data, effective saving factors and departure tendency corresponding to the employee samples and comprise: collecting the sample job leaving index data, the effective saving factors and the job leaving tendencies corresponding to the departed employees and the on-duty employees; the ratio of the number of the staff who leave the office to the number of the staff who are present at the office is a preset value; and preprocessing the sample departure index data, the effective saving factors and the departure tendency to generate the training sample.
Optionally, the preprocessing the sample departure index data comprises: performing data processing on the sample departure index data; wherein the data processing comprises: carrying out binarization processing on the discrete index data, and normalizing the continuous index data to a [0,1] interval; constructing a sample index feature vector based on the sample departure index data subjected to the data processing, and taking the sample index feature vector as the input quantity of the preset deep learning model.
Optionally, the pre-processing the effective saving factor and the tendency to leave employment comprises: setting the departure tendencies corresponding to the departed employees and the working employees to 1 and 0, respectively; setting an effective saving factor with the departed employee to 1.
Optionally, the sample job leaving index data and the predicted job leaving index data comprise: work cognition index data, value view index data and opportunity index data; wherein the work-recognition indicators include: department, employment property, job level, salary, working pressure, communication information receiving and transmitting amount, reward expectation and performance; the value indicators include: personal information, job level; the opportunity index comprises: employment opportunities.
Optionally, the sample index feature vector includes: and three sample index feature vectors respectively corresponding to the work cognition index, the value view index and the opportunity index data.
Optionally, a prediction index feature vector is constructed based on the predicted job leaving index data after the data processing, and the prediction index feature vector is used as an input quantity of the job leaving prediction model to obtain a job leaving tendency value and a saving factor strength of the predicted employee; wherein the predictor feature vector comprises: three prediction index feature vectors corresponding to the work cognition index, the value view index and the opportunity index data, respectively.
Optionally, the determining the tendency to leave and the factor to leave of the predicted employee according to the tendency to leave value and the saving factor strength comprises: if the value of the saving factor intensity is smaller than a preset intensity threshold value, determining the saving factor as the departure factor of the predicted employee; and if the value of the tendency to leave is larger than a preset tendency threshold value, determining that the predicted employee has the tendency to leave.
According to another aspect of the present disclosure, there is provided an employee departure prediction apparatus including: the sample generation module is used for generating training samples based on sample job leaving index data, effective saving factors and job leaving tendencies corresponding to the employee samples; the model training module is used for training a preset deep learning model by using a deep learning method and based on the training samples to obtain an out-of-position prediction model; the leave prediction module is used for updating the preset deep learning model into the leave prediction model, and obtaining the leave tendency value and the saving factor intensity of the predicted staff by inputting the predicted leave index data of the predicted staff into the leave prediction model; and the factor determining module is used for determining the departure tendency and the departure factor of the predicted employee according to the departure tendency value and the saving factor strength.
Optionally, the model training module is configured to use the sample job leaving index data as feature data, and use the effective saving factor and the job leaving tendency corresponding to the sample job leaving index data as an initial prediction result; and training the preset deep learning model based on the feature data and the initial prediction result.
Optionally, the preset deep learning model includes: a three-layer neuron model; the three-layer neuron model comprises: an input layer neuron model, a middle layer neuron model and an output layer neuron model; the output of each layer of neuron model is used as the input of the next layer of neuron model; wherein the neurons of the input layer neuron model correspond to the sample due indicators data, and the neurons of the output layer neuron model correspond to the retention factors and the tendency to cause a failure.
Optionally, the three-layer neuron model is a sub-network structure of a plurality of neural network layers having a fully connected structure; wherein the middle layer neuron model is a full-link layer.
Optionally, the employee sample comprises: employees who have left the job and employees who are on the job; the sample generation module is used for acquiring the sample departure index data, the effective saving factors and the departure tendency corresponding to the departed employees and the on-duty employees; the ratio of the number of the staff who leave the office to the number of the staff who are present at the office is a preset value; and preprocessing the sample departure index data, the effective saving factors and the departure tendency to generate the training sample.
Optionally, the sample generating module is configured to perform data processing on the sample departure index data; wherein the data processing comprises: carrying out binarization processing on the discrete index data, and normalizing the continuous index data to a [0,1] interval; constructing a sample index feature vector based on the sample departure index data subjected to the data processing, and taking the sample index feature vector as the input quantity of the preset deep learning model.
Optionally, the sample generating module is configured to set the tendency of leaving corresponding to the employee who has left the office and the employee who is on the office to 1 and 0, respectively; setting an effective saving factor with the departed employee to 1.
Optionally, the sample job leaving index data and the predicted job leaving index data comprise: work cognition index data, value view index data and opportunity index data; wherein the work-recognition indicators include: department, employment property, job level, salary, working pressure, communication information receiving and transmitting amount, reward expectation and performance; the value indicators include: personal information, job level; the opportunity index comprises: employment opportunities.
Optionally, the sample index feature vector includes: and three sample index feature vectors respectively corresponding to the work cognition index, the value view index and the opportunity index data.
Optionally, the leave prediction module is configured to construct a prediction index feature vector based on the predicted leave index data after the data processing, and use the prediction index feature vector as an input quantity of the leave prediction model, so as to obtain a leave tendency value and a leave factor strength of the predicted employee; wherein the predictor feature vector comprises: three prediction index feature vectors corresponding to the work cognition index, the value view index and the opportunity index data, respectively.
Optionally, the factor determining module is configured to determine the saving factor as the departure factor of the predicted employee if the value of the saving factor strength is smaller than a preset strength threshold; and if the value of the tendency to leave is larger than a preset tendency threshold value, determining that the predicted employee has the tendency to leave.
According to still another aspect of the present disclosure, there is provided an employee departure prediction apparatus including: a memory; and a processor coupled to the memory, the processor configured to perform the method as described above based on instructions stored in the memory.
According to yet another aspect of the present disclosure, a computer-readable storage medium is provided, which stores computer instructions for execution by a processor to perform the method as described above.
According to the employee leaving prediction method and device, a training sample is generated based on the leaving index data, the saving factor and the leaving tendency of an employee sample, a preset deep learning model is trained to obtain a leaving prediction model, and the leaving tendency value and the saving factor strength of a predicted employee are obtained by inputting the leaving index data of the predicted employee into the leaving prediction model; the employee job leaving tendency and the job leaving factors can be output at one time, employees with strong job leaving tendency can be predicted, and the human resource reservation strategy is combined, so that the human resource department can be effectively supported to screen and implement targeted reservation measures, the employee job leaving rate of an enterprise is remarkably reduced, the method is used for supporting the human loss and the human resource reservation measures, the human cost of the enterprise can be saved, and the normal operation or work progress of the enterprise can be guaranteed.
Drawings
In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present disclosure, and other drawings can be obtained by those skilled in the art without inventive exercise.
FIG. 1 is a schematic flow chart diagram illustrating one embodiment of a method for employee job departure prediction according to the present disclosure;
FIG. 2 is a schematic diagram of a deep learning model in one embodiment of an employee departure prediction method according to the present disclosure;
FIG. 3 is a block diagram view of one embodiment of an employee job departure prediction apparatus according to the present disclosure;
fig. 4 is a block diagram of another embodiment of an employee attendance prediction apparatus according to the present disclosure.
Detailed Description
The present disclosure now will be described more fully hereinafter with reference to the accompanying drawings, in which exemplary embodiments of the disclosure are shown. The technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the drawings in the embodiments of the present disclosure, and it is obvious that the described embodiments are only a part of the embodiments of the present disclosure, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments disclosed herein without making any creative effort, shall fall within the protection scope of the present disclosure. The technical solution of the present disclosure is described in various aspects below with reference to various figures and embodiments.
The terms "first", "second", and the like are used hereinafter only for descriptive distinction and not for other specific meanings.
Fig. 1 is a flowchart illustrating an embodiment of an employee departure prediction method according to the present disclosure, and as shown in fig. 1, the employee departure prediction method includes steps 101-104.
Step 101, generating training samples based on sample departure index data corresponding to the employee samples, effective saving factors and departure tendencies.
And 102, training a preset deep learning model by using a deep learning method based on the training samples to obtain an out-of-position prediction model.
And 103, updating the preset deep learning model into a departure prediction model, and inputting the predicted departure index data of the predicted staff into the departure prediction model to obtain the departure tendency value and the saving factor strength of the predicted staff.
And step 104, determining the departure tendency and the departure factor of the predicted employee according to the departure tendency value and the saving factor strength. The job leaving prediction model can output the job leaving tendency and the job leaving factors of the staff at one time, and is combined with a human resource reservation strategy to support human loss and human resource reservation measures.
In one embodiment, the employee is paying the effort and the company is awarded the reward. From the company's perspective, the company gives a certain reward to save the employee to obtain a constant payment from the employee, and the company gives the employee a saving that can be considered. The satisfaction degree of the employee on the saving measures of the company is high, and the employee will be weak to leave; otherwise, the willingness of the staff to leave the job is strong. The saving factor can be company controllable factors (such as salary, job level and the like), and the intensity of willingness of staff to leave can be weakened by adjusting the factors. The retention factor intensity refers to the self-awareness of the degree of satisfaction that an employee achieves in terms of the job leaving retention factor. Work cognition, value view, work change opportunity and the like are important influence factors of employee job leaving, and the job leaving tendency can reflect the job leaving tendency of the employee. The retention factor intensity may reflect the employee's self-awareness of the retention factor intensity.
The sample departure index data and the predicted departure index data have the same content, and can comprise work cognition index data, value view index data, opportunity index data and the like. The work cognitive index can include departments, employment properties, job levels, salaries, work pressure, communication information receiving and sending amount, reward expectation, performance and the like. The value view indicators may include personal information, job title, and the like. Opportunity metrics may include employment opportunities, and the like.
The withholding factor refers to a quantitative factor that can be adjusted by a company for an individual employee, such as job level (adjusting job level of the employee), salary (adjusting salary treatment of the employee), department (adjusting organization architecture to which the employee belongs), and the like. The saving factors may be a specific set of factors (job level, compensation, department), and the set may intersect with the total set of the work-recognition index, the value observation index, and the opportunity index, and is not necessarily an inclusion relationship.
In the aspect of employee departure willingness analysis, the analysis models mainly include Marchimeng models, Price models, Mobely intermediate chain models, extended Mobely models and the like. The Mobely model is expanded to be used as the synthesis of the first three models, so that better explanatory power is achieved. And modifying the extended Mobely model as a basic model, adding the influence on the intensity of the saving factor on the basic model in order to analyze the intensity of the saving factor, slightly simplifying the complex basic model, and preferably selecting reasonable factors as input. The remaining factors may be set as compensation, job level, post, nature of the job, etc. The employee departure index system is shown in the following table 1, and the saving factors, the departure indexes, the index names, the specific indexes and the like can be adjusted and are not limited to the enumeration range.
TABLE 1 staff leave index System
In one embodiment, sample departure index data is used as feature data, effective saving factors and departure tendencies corresponding to the sample departure index data are used as initial prediction results, one sample departure index data corresponds to one group of effective saving factors and departure tendencies, and a preset deep learning model is trained based on the feature data and the initial prediction results. There are various deep learning models, for example, deep learning models include CNN, DBN, RNN, RNTN, auto-encoder, GAN, and the like. The preset deep learning model comprises three layers of neuron models, each three layer of neuron model comprises an input layer neuron model, a middle layer neuron model and an output layer neuron model, the output of each layer of neuron model is used as the input of the next layer of neuron model, the neurons of the input layer neuron model correspond to sample departure index data, and the neurons of the output layer neuron model correspond to a saving factor and a departure tendency. The three-layer neuron model may be a sub-network structure having a plurality of neural network layers of a fully-connected structure, and the middle-layer neuron model is a fully-connected layer.
The employee samples include: employees who are out of office and employees who are in office. Collecting sample departure index data, effective saving factors, departure tendency and the like corresponding to the employees who leave the job and the employees who are in the job; the ratio of the number of the staff who leave to the number of the staff who are present is a preset value. When any employee submits a departure, the HR is asked to ask the departing employee to answer the reason for the departure by way of a personal query or questionnaire. When the reason for employee job leaving includes the saving factor, the saving case is marked by the saving factor, and one saving case can mark a plurality of perception saving factors. In the existing human resource system, the information acquisition is basically carried out, so that the partial content of data acquisition can not stress the existing human resource system.
And preprocessing the sample job leaving index data, the effective saving factors and the job leaving tendency to generate a training sample. There are various methods for preprocessing the sample job leaving index data. For example, the sample departure index data is subjected to data processing, and the data processing comprises the following steps: and (3) carrying out binarization processing on discrete index data, normalizing continuous index data to a [0,1] interval and the like. And constructing a sample index feature vector based on the sample departure index data subjected to data processing, and taking the sample index feature vector as the input quantity of a preset deep learning model.
There are several methods for pre-treating effective retaining factors and job departure tendencies. For example, the tendency to leave corresponding to the employee who has left and the employee who is present are set to 1 and 0, respectively, and the effective retention factor with the employee who has left is set to 1. The sample index feature vectors include three sample index feature vectors corresponding to the work cognitive index, the value view index, and the opportunity index data, respectively. And constructing a prediction index feature vector based on the predicted job leaving index data subjected to data processing, and taking the prediction index feature vector as an input quantity of a job leaving prediction model for obtaining a job leaving tendency value and a saving factor strength of the predicted staff. The prediction index feature vector comprises three prediction index feature vectors respectively corresponding to the work cognition index, the value view index and the opportunity index data.
For example, each item of data of the existing human resource system includes personal information, work information, compensation information, position information and the like of the staff supporting staff departure warning. Along with the flow of company personnel, the human resource data is continuously enriched, and a more accurate model can be trained. And reading data information of the staff from a human resource database, wherein the data information comprises two parts of the staff who leave and the staff who do not leave, and the ratio is 1: 3. For the employees who have already left the job, the read data comprises various job leaving index data and saving factors, and the supplemented job leaving tendency is 1. For employees who do not leave the job, the read data comprises various leaving index data, various saving factors are supplemented to be 1, one or more of the leaving index data can be set as the saving factors, and the tendency to leave the job is set to be 0. And forming a training sample set according to the acquired data information of the departed staff and the data information of the non-departed staff.
In one embodiment, a deep learning backbone network structure is designed according to characteristics of staff departure warning application in the aspects of available data, domain prior knowledge, output results and the like, a sub-network structure for outputting the intensity (influence degree) of a retention factor is embedded, and an end-to-end network is constructed. The deep learning network structure for employee departure prediction application has the characteristics of universality, objectivity, robustness and end-to-end performance.
As shown in fig. 2, Output 1x4 represents the persistence factor strength Output, with a value range of 0, 1. If the intensity of the specific saving factor is less than 0.5, the saving factor belongs to a main factor for staff leaving, otherwise, the lower the value is, the weaker the intensity of the specific saving factor is, and the stronger the effect of adjusting the factor to save the staff is. Output 1x1 represents the departure tendency Output, and the value range is [0,1], and the higher the value is, the stronger the departure tendency is. The Input1-3 corresponds to the specific index Input of three factor structures of work cognition, value view and opportunity respectively. The input factors are the specific indexes enumerated in table 1.
Units 1-5 correspond to a deep learning network substructure respectively. The sub-structure design and complexity of the network are different according to the complexity of the problem. Unit 1-5 can be designed as a single neural network layer with a certain number of neuron nodes. For example, Unit 1-5 can be designed as a neural network layer with 50 neurons. When the precision of a single layer of neurons is low, one or more of units 1-5 can be designed as a sub-network structure with multiple neural network layers in a fully connected structure.
The training sample set is processed by data preprocessing and data quantization. The data preprocessing refers to binarizing each discrete index, normalizing each continuous index to a [0,1] interval, and the data quantization refers to splicing each index of each factor into a vector. Through preprocessing and tensor operation, each piece of input data corresponds to three vectors of work cognition, value view and opportunity. The data vectors are input into a deep learning model for training, and the training framework and parameters are not limited. Through training, the departure early warning model can be obtained, and the trained departure early warning model can be updated on line at regular time or according to the leisure time of the prediction module.
When the departure prediction is carried out, the departure tendency and the saving factor strength of a specific employee are predicted according to the information of the employee. The departure prediction can be performed periodically or on demand according to the human resource business requirements, and the prediction subdivision unit is an individual, so that analysis can be performed according to departments, career departments and crowds. In single execution, each input employee information is subjected to data preprocessing and data tensor quantization, the obtained three tensors are input into a deep learning engine loaded with a leaving prediction model, the leaving prediction model outputs the leaving tendency and the saving factor strength of the employee, and the output information is stored in a database of a human resource system and used for subsequent analysis.
If the value of the stay factor intensity is less than the preset intensity threshold, the stay factor is determined as the departure factor of the predicted employee. For example, the preset intensity threshold may be 0.3, and if the value of the intensity of the saving factor output by the leaving prediction model is less than 0.3, the saving factor is determined as the leaving factor of the predicted employee. And if the value of the tendency to leave is larger than a preset tendency threshold value, determining that the predicted employee has the tendency to leave. For example, the preset tendency threshold may be 0.6, and if the value of the tendency to leave output by the leave prediction model is greater than 0.6, the predicted employee is determined to have a tendency to leave.
In one embodiment, the present disclosure provides an employee departure prediction apparatus 30, comprising: a sample generation module 31, a model training module 32, an escape prediction module 33, and a factor determination module 34. The sample generation module 31 generates training samples based on sample departure index data, effective retention factors, and tendency to depart corresponding to the employee samples. The model training module 32 trains a preset deep learning model by using a deep learning method based on training samples to obtain an out-of-position prediction model.
The leaving prediction module 33 updates the preset deep learning model to a leaving prediction model, and obtains the leaving tendency value and the saving factor strength of the predicted employee by inputting the predicted leaving index data of the predicted employee into the leaving prediction model. The factor determination module 34 determines the departure propensity and the departure factor of the predicted employee based on the departure propensity value and the strength of the retention factor.
In one embodiment, the model training module 32 takes the sample job leaving index data as the feature data, takes the effective saving factor and the job leaving tendency corresponding to the sample job leaving index data as the initial prediction result, and trains the preset deep learning model based on the feature data and the initial prediction result. The preset deep learning model comprises the following steps: three-layer neuron model. The three-layer neuron model comprises: an input layer neuron model, a middle layer neuron model, and an output layer neuron model. The output of each layer of neuron model is used as the input of the next layer of neuron model, the neurons of the input layer of neuron model correspond to sample departure index data, and the neurons of the output layer of neuron model correspond to a saving factor and a departure tendency. The three-layer neuron model is a sub-network structure with a plurality of neural network layers of a full connection structure, and the middle-layer neuron model is a full connection layer.
In one embodiment, the employee samples include employees who have left the job and employees who are on the job. The sample generation module 31 collects sample departure index data, effective saving factors and departure tendencies corresponding to the employees who have left the job and the employees who are in the job, and the ratio of the number of the employees who have left the job to the number of the employees who are in the job is a preset value. The sample generation module 31 preprocesses the sample departure index data, the effective saving factors and the departure tendency to generate a training sample.
The sample generation module 31 performs data processing on the sample departure index data, and the data processing includes: and (3) carrying out binarization processing on discrete index data, normalizing continuous index data to a [0,1] interval and the like. The sample generation module 31 constructs a sample index feature vector based on the sample job leaving index data after data processing, and uses the sample index feature vector as an input quantity of a preset deep learning model. The sample generation module 31 sets the tendency to leave corresponding to the employee who has left and the employee who is present to 1 and 0, respectively, and sets the effective saving factor corresponding to the employee who has left and the employee who is present to 1.
The leave prediction module 33 constructs a prediction index feature vector based on the predicted leave index data after data processing, and uses the prediction index feature vector as an input quantity of a leave prediction model, so as to obtain a leave tendency value and a saving factor strength of the predicted employee. The prediction index feature vector comprises three prediction index feature vectors respectively corresponding to the work cognition index, the value view index and the opportunity index data. If the value of the stay factor intensity is less than the preset intensity threshold, the factor determination module 34 determines the stay factor as the departure factor of the predicted employee, and if the value of the departure tendency is greater than the preset tendency threshold, the factor determination module 34 determines that the predicted employee has a departure tendency.
Fig. 4 is a block diagram of another embodiment of an employee attendance prediction apparatus according to the present disclosure. As shown in fig. 4, the apparatus may include a memory 41, a processor 42, a communication interface 43, and a bus 44. The memory 41 is used for storing instructions, the processor 42 is coupled to the memory 41, and the processor 42 is configured to execute the employee attendance prediction method described above based on the instructions stored by the memory 41.
The memory 41 may be a high-speed RAM memory, a non-volatile memory (non-volatile memory), or the like, and the memory 41 may be a memory array. The storage 41 may also be partitioned, and the blocks may be combined into virtual volumes according to certain rules. Processor 42 may be a central processing unit CPU, or an application specific integrated circuit asic (application specific integrated circuit), or one or more integrated circuits configured to implement the employee job departure prediction method of the present disclosure.
In one embodiment, the present disclosure provides a computer-readable storage medium having stored thereon computer instructions that, when executed by a processor, implement an employee departure prediction method as in any one of the above embodiments.
According to the employee leaving prediction method and device in the embodiment, the training sample is generated based on the leaving index data, the saving factor and the leaving tendency of the employee sample, the preset deep learning model is trained to obtain the leaving prediction model, and the leaving tendency value and the saving factor strength of the predicted employee are obtained by inputting the leaving index data of the predicted employee into the leaving prediction model; the employee job leaving tendency and the job leaving factors can be output at one time, employees with strong job leaving tendency can be predicted, and the human resource reservation strategy is combined, so that the human resource department can be effectively supported to screen and implement targeted reservation measures, the employee job leaving rate of an enterprise is remarkably reduced, the method is used for supporting the human loss and the human resource reservation measures, the human cost of the enterprise can be saved, and the normal operation or work progress of the enterprise can be guaranteed.
The method and system of the present disclosure may be implemented in a number of ways. For example, the methods and systems of the present disclosure may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above-described order for the steps of the method is for illustration only, and the steps of the method of the present disclosure are not limited to the order specifically described above unless specifically stated otherwise. Further, in some embodiments, the present disclosure may also be embodied as programs recorded in a recording medium, the programs including machine-readable instructions for implementing the methods according to the present disclosure. Thus, the present disclosure also covers a recording medium storing a program for executing the method according to the present disclosure.
The description of the present disclosure has been presented for purposes of illustration and description, and is not intended to be exhaustive or limited to the disclosure in the form disclosed. Many modifications and variations will be apparent to practitioners skilled in this art. The embodiment was chosen and described in order to best explain the principles of the disclosure and the practical application, and to enable others of ordinary skill in the art to understand the disclosure for various embodiments with various modifications as are suited to the particular use contemplated.
Claims (24)
1. An employee departure prediction method comprising:
generating a training sample based on sample job leaving index data, effective saving factors and job leaving tendencies corresponding to the employee sample;
training a preset deep learning model by using a deep learning method based on the training samples to obtain an out-of-position prediction model;
updating the preset deep learning model into the departure prediction model, and inputting predicted departure index data of the predicted staff into the departure prediction model to obtain the departure tendency value and the saving factor strength of the predicted staff;
and determining the departure tendency and the departure factor of the predicted employee according to the departure tendency value and the saving factor strength.
2. The method of claim 1, the training a preset deep learning model based on the training samples comprising:
taking the sample job leaving index data as feature data, and taking the effective saving factor and the job leaving tendency corresponding to the sample job leaving index data as initial prediction results;
and training the preset deep learning model based on the feature data and the initial prediction result.
3. The method of claim 2, wherein,
the preset deep learning model comprises: a three-layer neuron model; the three-layer neuron model comprises: an input layer neuron model, a middle layer neuron model and an output layer neuron model; the output of each layer of neuron model is used as the input of the next layer of neuron model;
wherein the neurons of the input layer neuron model correspond to the sample due indicators data, and the neurons of the output layer neuron model correspond to the retention factors and the tendency to cause a failure.
4. The method of claim 3, wherein,
the three-layer neuron model is a sub-network structure with a plurality of neural network layers of a full connection structure; wherein the middle layer neuron model is a full-link layer.
5. The method of claim 1, the employee sample comprising: the training samples are generated based on sample departure index data, effective saving factors and departure tendency corresponding to the employee samples and comprise:
collecting the sample job leaving index data, the effective saving factors and the job leaving tendencies corresponding to the departed employees and the on-duty employees; the ratio of the number of the staff who leave the office to the number of the staff who are present at the office is a preset value;
and preprocessing the sample departure index data, the effective saving factors and the departure tendency to generate the training sample.
6. The method of claim 5, the preprocessing the sample due indicator data comprising:
performing data processing on the sample departure index data; wherein the data processing comprises: carrying out binarization processing on the discrete index data, and normalizing the continuous index data to a [0,1] interval;
constructing a sample index feature vector based on the sample departure index data subjected to the data processing, and taking the sample index feature vector as the input quantity of the preset deep learning model.
7. The method of claim 6, wherein said pre-processing said effective saving factors, said tendency to leave employment comprises:
setting the departure tendencies corresponding to the departed employees and the working employees to 1 and 0, respectively;
setting an effective saving factor corresponding to the departed employee to 1.
8. The method of claim 7, wherein,
the sample job leaving indicator data and the predicted job leaving indicator data comprise: work cognition index data, value view index data and opportunity index data;
wherein the work-recognition indicators include: department, employment property, job level, salary, working pressure, communication information receiving and transmitting amount, reward expectation and performance; the value indicators include: personal information, job level; the opportunity index comprises: employment opportunities.
9. The method of claim 8, wherein,
the sample index feature vector includes: and three sample index feature vectors respectively corresponding to the work cognition index, the value view index and the opportunity index data.
10. The method of claim 8, wherein,
constructing a prediction index feature vector based on the predicted job leaving index data subjected to the data processing, and taking the prediction index feature vector as an input quantity of the job leaving prediction model to obtain a job leaving tendency value and a saving factor strength of the predicted employee;
wherein the predictor feature vector comprises: three prediction index feature vectors corresponding to the work cognition index, the value view index and the opportunity index data, respectively.
11. The method of claim 1, the determining the departure tendency and the departure factor for the predicted employee based on the departure tendency value and the saving factor strength comprising:
if the value of the saving factor intensity is smaller than a preset intensity threshold value, determining the saving factor as the departure factor of the predicted employee;
and if the value of the tendency to leave is larger than a preset tendency threshold value, determining that the predicted employee has the tendency to leave.
12. An employee departure prediction apparatus comprising:
the sample generation module is used for generating training samples based on sample job leaving index data, effective saving factors and job leaving tendencies corresponding to the employee samples;
the model training module is used for training a preset deep learning model by using a deep learning method and based on the training samples to obtain an out-of-position prediction model;
the leave prediction module is used for updating the preset deep learning model into the leave prediction model, and obtaining the leave tendency value and the saving factor intensity of the predicted staff by inputting the predicted leave index data of the predicted staff into the leave prediction model;
and the factor determining module is used for determining the departure tendency and the departure factor of the predicted employee according to the departure tendency value and the saving factor strength.
13. The apparatus of claim 12, wherein,
the model training module is used for taking the sample job leaving index data as feature data and taking the effective saving factor and the job leaving tendency corresponding to the sample job leaving index data as initial prediction results; and training the preset deep learning model based on the feature data and the initial prediction result.
14. The apparatus of claim 13, wherein,
the preset deep learning model comprises: a three-layer neuron model; the three-layer neuron model comprises: an input layer neuron model, a middle layer neuron model and an output layer neuron model; the output of each layer of neuron model is used as the input of the next layer of neuron model;
wherein the neurons of the input layer neuron model correspond to the sample due indicators data, and the neurons of the output layer neuron model correspond to the retention factors and the tendency to cause a failure.
15. The apparatus of claim 14, wherein,
the three-layer neuron model is a sub-network structure with a plurality of neural network layers of a full connection structure; wherein the middle layer neuron model is a full-link layer.
16. The apparatus of claim 12, the employee sample comprising: employees who have left the job and employees who are on the job;
the sample generation module is used for acquiring the sample departure index data, the effective saving factors and the departure tendency corresponding to the departed employees and the on-duty employees; the ratio of the number of the staff who leave the office to the number of the staff who are present at the office is a preset value; and preprocessing the sample departure index data, the effective saving factors and the departure tendency to generate the training sample.
17. The apparatus of claim 16, wherein,
the sample generation module is used for carrying out data processing on the sample departure index data; wherein the data processing comprises: carrying out binarization processing on the discrete index data, and normalizing the continuous index data to a [0,1] interval; constructing a sample index feature vector based on the sample departure index data subjected to the data processing, and taking the sample index feature vector as the input quantity of the preset deep learning model.
18. The apparatus of claim 17, wherein,
the sample generation module is used for respectively setting the departure tendency corresponding to the departed employee and the employee to be 1 and 0; setting an effective saving factor with the departed employee to 1.
19. The apparatus of claim 18, wherein,
the sample job leaving indicator data and the predicted job leaving indicator data comprise: work cognition index data, value view index data and opportunity index data;
wherein the work-recognition indicators include: department, employment property, job level, salary, working pressure, communication information receiving and transmitting amount, reward expectation and performance; the value indicators include: personal information, job level; the opportunity index comprises: employment opportunities.
20. The apparatus of claim 19, wherein,
the sample index feature vector includes: and three sample index feature vectors respectively corresponding to the work cognition index, the value view index and the opportunity index data.
21. The apparatus of claim 19, wherein,
the job leaving prediction module is used for constructing a prediction index feature vector based on the predicted job leaving index data subjected to the data processing, and taking the prediction index feature vector as an input quantity of the job leaving prediction model to obtain a job leaving tendency value and a saving factor strength of the predicted employee; wherein the predictor feature vector comprises: three prediction index feature vectors corresponding to the work cognition index, the value view index and the opportunity index data, respectively.
22. The apparatus of claim 12, wherein,
the factor determining module is used for determining the saving factor as the departure factor of the predicted employee if the value of the saving factor intensity is smaller than a preset intensity threshold value; and if the value of the tendency to leave is larger than a preset tendency threshold value, determining that the predicted employee has the tendency to leave.
23. An employee departure prediction apparatus comprising:
a memory; and a processor coupled to the memory, the processor configured to perform the method of any of claims 1-11 based on instructions stored in the memory.
24. A computer-readable storage medium having stored thereon computer instructions for execution by a processor to perform the method of any one of claims 1 to 11.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810696834.9A CN110659757A (en) | 2018-06-29 | 2018-06-29 | Employee departure prediction method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810696834.9A CN110659757A (en) | 2018-06-29 | 2018-06-29 | Employee departure prediction method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110659757A true CN110659757A (en) | 2020-01-07 |
Family
ID=69026698
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810696834.9A Withdrawn CN110659757A (en) | 2018-06-29 | 2018-06-29 | Employee departure prediction method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110659757A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111429114A (en) * | 2020-04-16 | 2020-07-17 | 上海应用技术大学 | Staff loss early warning pre-control mechanism system |
| CN111709714A (en) * | 2020-06-17 | 2020-09-25 | 腾讯云计算（北京）有限责任公司 | Method and device for predicting lost personnel based on artificial intelligence |
| WO2021179715A1 (en) * | 2020-10-21 | 2021-09-16 | 平安科技（深圳）有限公司 | Hidden markov model-based resignation prediction method and related device |
| CN113706013A (en) * | 2021-08-27 | 2021-11-26 | 上海见兴信息科技有限公司 | Labor relation contradiction risk analysis method combining financial technical indexes |
-
2018
- 2018-06-29 CN CN201810696834.9A patent/CN110659757A/en not_active Withdrawn
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111429114A (en) * | 2020-04-16 | 2020-07-17 | 上海应用技术大学 | Staff loss early warning pre-control mechanism system |
| CN111709714A (en) * | 2020-06-17 | 2020-09-25 | 腾讯云计算（北京）有限责任公司 | Method and device for predicting lost personnel based on artificial intelligence |
| CN111709714B (en) * | 2020-06-17 | 2024-03-29 | 腾讯云计算（北京）有限责任公司 | Loss personnel prediction method and device based on artificial intelligence |
| WO2021179715A1 (en) * | 2020-10-21 | 2021-09-16 | 平安科技（深圳）有限公司 | Hidden markov model-based resignation prediction method and related device |
| CN113706013A (en) * | 2021-08-27 | 2021-11-26 | 上海见兴信息科技有限公司 | Labor relation contradiction risk analysis method combining financial technical indexes |
| CN113706013B (en) * | 2021-08-27 | 2023-12-29 | 上海见兴信息科技有限公司 | Labor relation contradiction risk analysis method combined with financial technical index |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Srivastava et al. | Intelligent employee retention system for attrition rate analysis and churn prediction: An ensemble machine learning and multi-criteria decision-making approach | |
| CN110659757A (en) | Employee departure prediction method and device | |
| Antonsen | The downside of the Balanced Scorecard: A case study from Norway | |
| CN111401828A (en) | Dynamic intelligent interviewing method, device and equipment for strengthening sorting and computer storage medium | |
| Stern et al. | Decision making, procedural compliance, and outcomes definition in US forest service planning processes | |
| Steiber et al. | Trends in work stress and exhaustion in advanced economies | |
| CN112036715A (en) | Regional human resource development index measuring system and method | |
| CN112926943A (en) | Automatic scheduling method, system, computer equipment and storage medium | |
| CN113947322A (en) | Figure matching method and system based on FP-Growth algorithm | |
| Brightenburg et al. | Job engagement levels across the generations at work | |
| Medina-Garrido et al. | Organizational support for work-family life balance as an antecedent to the well-being of tourism employees in Spain | |
| Fahrizal et al. | Unlocking Work Engagement: How Leadership and Total Rewards Impact Employee Work Engagement Through the Mediating Role of Service Climate in Supply Chain and Logistic Company in Indonesia | |
| CN113435713B (en) | Risk map compiling method and system based on GIS technology and two-model fusion | |
| JP2020160551A (en) | Analysis support device for personnel item, analysis support method, program, and recording medium | |
| Kotni | E-Hrm-a Paradigm Shift in Hr Practices and Its Effects on Perception of Employees Towards Accepting This New Technology | |
| Chen et al. | Extended DEA model under type-2 fuzzy environment: An application of rural poverty reduction in Hainan province | |
| Bačová et al. | Too far away to care about?: predicting psychological preparedness for retirement financial planning among young employed adults | |
| CN113408930B (en) | Employee behavior analysis method, system and storage medium based on employee service system | |
| Dubey et al. | A theoretical framework of soft TQM in successful implementation | |
| Cornelissen et al. | Profit sharing and reciprocity: Theory and survey evidence | |
| Coelho et al. | ANTECEDENT DETERMINANTS OF BELIEFS UNDERPINNING THE INTENTION TO ACCEPT AND USE BUSINESS INTELLIGENCE SYSTEMS: THE IMPORTANCE OF ORGANIZATIONAL FACTORS. | |
| Algawazi et al. | The impact of the organizational culture traits on task performance through personality traits among the employees of the semi-governmental organisations in Saudi Arabia | |
| Kulchitskaya et al. | Cognitive and motivational aspects of HR audit implementation in an organization | |
| Bowlin | Benchmarking and efficient program delivery for the Department of Defense's business-like activities | |
| Jiang-tao et al. | The application of the analytical hierarchy process in performance evaluation system in commercial bank's IT department |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20200107 |