Human resource data classification statistical method and system based on big data

Abstract

The invention discloses a human resource data classification statistical method and system based on big data. The method comprises the following steps: collecting employee basic information and employee behavior data; performing coding processing on the staff statistical data, establishing a staff multi-modal feature set according to the staff basic information and the staff behavior data, inputting the staff multi-modal feature set into a deep learning model, and predicting staff skill scores; extracting the employee behavior data and the employee skill score, constructing an employee behavior feature matrix and a post skill feature matrix, setting a first potential factor and predicting an employee stable value according to the behavior feature matrix, and setting a second potential factor and predicting a post matching value according to the post skill feature matrix; according to the employee stable value and the post matching degree, post recommendation scores are distributed, and a recommendation score thermodynamic density map is visually output; according to the method, the whole process from data acquisition to decision recommendation is automatic, talent evaluation precision is realized, employee demission risks are effectively predicted, and digital transformation of human resource management is realized.

Description

Technical Field

[0001] The present invention relates to the field of human resource classification and big data technology, and in particular to a human resource data classification and statistics method and system based on big data. Background Art

[0002] With the rise of artificial intelligence, many companies have ushered in a wave of digital transformation. Human resource management has become an important issue for enterprises and society. Traditional human resource statistical management relies solely on manual classification statistics and Excel spreadsheet analysis, which is inefficient and consumes labor costs.

[0003] The current recruitment and human resources data management fields mainly adopt simple manual statistics and database retrieval methods to match job requirements and analyze employee training programs and development directions. They fail to fully analyze employee work information and provide career development directions, resulting in poor job stability for employees. Therefore, big data analysis is urgently needed to help companies intelligently analyze the capabilities and development needs of current employees, optimize the company's human resources management model, and provide decision-making support for management. Summary of the Invention

[0004] The technical problem solved by the present invention is to effectively generate job recommendations based on employee skill characteristics, achieve rapid adaptation of employees to jobs, and optimize the enterprise human resource management method.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: In the first aspect, a human resources data classification and statistics method based on big data comprises: Step S1: Collect employee basic information and employee behavior data and save them as employee statistics; Step S2: performing label encoding and one-hot encoding on the employee statistical data, establishing a multimodal feature set of employees based on the employee basic information and employee behavior data, inputting the multimodal feature set of employees into a deep learning model, and predicting employee skill scores; Step S3: extract the employee behavior data and employee skill scores, construct an employee behavior feature matrix and a job skill feature matrix, set a first latent factor based on the behavior feature matrix, predict the employee stability value, set a second latent factor based on the job skill feature matrix, and predict the job matching value.

[0006] Step S4: assigning a job recommendation score based on the employee stability value and the job matching degree, and visually outputting a heat density map of the recommendation score.

[0007] Preferably, the basic employee information includes: employee number, age, years of service, education, gender, marital status and department, and the employee behavior data includes: employee number, attendance records, historical performance scores, training participation, job satisfaction and team contribution value.

[0008] Preferably, step S1 specifically includes: Collect employee basic information, use the employee number as a unique index, and save it in the database. Obtain employee behavior data based on the company's employee system, use the employee number as a common index, and use the employee behavior data and employee basic information as employee statistics.

[0009] Preferably, step S2 specifically includes: The education level and job satisfaction in the employee basic information and employee behavior data are coded according to the level of education level and job satisfaction from high to low, and the gender, marital status and department are coded using a one-hot encoding method. The employee basic information and employee behavior data are saved as a numerical feature matrix, a categorical feature matrix, a text feature matrix, and a time series feature matrix according to the numerical type, categorical type, text type, and time series type, respectively, and all feature matrices are merged to form a multimodal feature set; A deep learning network is constructed using BERT and LSTM network settings. The multimodal feature set is extracted and enhanced through different neural network branches. The numerical features are processed using a fully connected layer activated by RELU. Dropout is used to prevent overfitting of the categorical features. The text features are stably trained through LayerNorm. The temporal features are activated by ELU to capture nonlinear relationships. The weights of each feature are dynamically calculated and fused through a cross-modal attention mechanism to form a unified feature representation. A multi-layer fully connected network is used to predict and output the skill score.

[0010] Preferably, step S3 specifically includes: a feature matrix module, an employee stability value prediction module and an employee position matching module; The feature matrix module includes: calculating the attendance abnormality rate based on the attendance records of the employee behavior data ,in The abnormal attendance rate of the nth employee is the ratio of abnormal attendance days to the total attendance days. The standard deviation of performance fluctuations for each quarter is calculated based on the historical performance score. , based on the training participation rate, count the number of training sessions that employees have taken in the past three months , using the attendance abnormality rate, performance fluctuation standard deviation and training times to construct the employee behavior characteristic matrix , construct a job skill feature matrix based on the skill score x, the team contribution value y and the job satisfaction z .

[0011] Preferably, the employee stability value prediction module includes: using NFM to decompose the employee behavior feature matrix to obtain an employee engagement factor and a risk propensity factor, the sum of the employee engagement factor and the risk propensity factor being equal to 1; using a logistic regression model to perform model stability training on the employee engagement factor and the risk propensity factor to obtain a first weight factor and a second weight factor, wherein the first weight factor corresponds to the employee engagement factor and the second weight factor corresponds to the risk propensity factor; calculating a first latent factor, and the calculation expression of the first latent factor m is: ; According to the first potential factor, the employee stability value is calculated. The calculation expression of the employee stability value is: ; Among them, p is the employee stability value, e is a mathematical constant, and m is the first potential factor.

[0012] Preferably, the employee position matching module includes: using NFM to decompose the position skill feature matrix to obtain a technical ability factor and a management ability factor, the sum of the technical ability factor and the management ability factor is equal to 1, using cosine similarity to perform model stability training on the technical ability factor and the management ability factor to obtain a third weight factor and a fourth weight factor, wherein the third weight factor corresponds to the technical ability factor, and the fourth weight factor corresponds to the management ability factor, and calculating a second latent factor, and the calculation expression of the second latent factor k is: ; According to the second potential factor, the position matching value is calculated. The calculation expression of the position matching value is: ; Among them, l is the job matching value, e is a mathematical constant, and k is the second potential factor.

[0013] Preferably, step S4 specifically includes: Manually set the distribution histogram of the recommendation score, use the employee stability value as the horizontal distribution value, and the job matching degree as the vertical distribution value, obtain the employee job matching stability coordinates, and map them into the distribution histogram. According to the distribution position of the employee job matching stability coordinates, match the corresponding scores to obtain the job recommendation scores, use K-means density clustering to correspond the job recommendation scores to the employee job matching stability coordinates one by one, and visualize the output to obtain a large screen of employee evaluation density clustering.

[0014] Preferably, the employee evaluation density clustering screen is also used to query the current employee's on-the-job status, job score prediction and on-the-job event trajectory based on the employee number and department, and to query the number of employees in the current department, department stability and on-the-job distribution of department personnel based on the department.

[0015] Secondly, a human resources data classification and statistics system based on big data, including a data collection module, a skill scoring module, a characteristic value prediction module and a visual analysis module; The data collection module is used to collect basic employee information and employee behavior data and save them in the corresponding employee log; The skill scoring module is used to perform label encoding and one-hot encoding on the employee statistical data, classify the employee basic information and employee behavior data according to their data types, and establish a multimodal feature set for the employees. The multimodal feature set is input into the deep learning model to predict the employee skill score; The characteristic value prediction module is used to extract the employee behavior data and employee skill scores, sort the data into an employee behavior feature matrix and a job skill feature matrix, set a first latent factor based on the behavior feature matrix, predict the employee stability value, and set a second latent factor based on the job skill feature matrix to predict the job matching value.

[0016] The visualization analysis module is used to assign job recommendation scores according to the employee stability value and job matching degree, and to visually output a heat density map of the recommendation scores.

[0017] The beneficial effects of the present invention are as follows: by collecting basic employee information and employee behavior data, comprehensively monitoring the distribution of employee behavior data on the job, and effectively using the deep learning network to predict employee skill scores based on the data, this method automatically extracts feature associations through deep learning, avoiding the risk of inconsistent judgment accuracy caused by human subjective judgment. Secondly, based on the job matching calculation of the employees, the matching value between the employees and the jobs can be effectively quantified, effectively reducing the job adjustment cycle and improving the production efficiency of the enterprise. Finally, based on k-means density clustering and creating a large visual screen, employee behavior events can be visualized, and job recommendations can be effectively generated based on employee skill characteristics, thereby achieving rapid adaptation of employees and jobs. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic diagram of the basic flow of a human resources data classification and statistics method based on big data provided by one embodiment of the present invention.

[0019] Figure 2 A schematic diagram of the basic flow of a human resources data classification and statistics system based on big data is provided for one embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0021] Example 1, with reference to Figure 1 , as an embodiment of the present invention, provides a human resources data classification and statistics method based on big data, comprising: Step S1: Collect employee basic information and employee behavior data and save them as employee statistics; Step S2: Label encoding and one-hot encoding are performed on employee statistical data. A multimodal feature set of employees is established based on their basic information and behavioral data. The multimodal feature set of employees is input into a deep learning model to predict employee skill scores. Step S3: extract employee behavior data and employee skill scores, construct an employee behavior feature matrix and a job skill feature matrix, set a first latent factor based on the behavior feature matrix, predict the employee stability value, set a second latent factor based on the job skill feature matrix, and predict the job matching value.

[0022] Step S4: assign job recommendation scores based on employee stability values ​​and job matching degrees, and visualize and output a heat density map of the recommendation scores.

[0023] In this embodiment, the implementation of the method is mainly divided into four steps, namely data collection, feature engineering construction, matrix factor prediction and large-screen visualization. By integrating multiple modules of the collected employee data, a deep learning network model is intelligently constructed to predict employee skill scores, and a multi-dimensional feature matrix is ​​constructed. Based on the employee's behavioral characteristics and job skills, the stability value and matching value are predicted, realizing the precision and digitization of human resource management, and effectively solving the judgment bias caused by human judgment.

[0024] Employee basic information includes: employee number, age, years of service, education, gender, marital status and department; employee behavior data includes: employee number, attendance records, historical performance scores, training participation, job satisfaction and team contribution value.

[0025] In this embodiment, the specific contents of employee basic information and employee behavior data are specifically explained. Seven basic information items and six types of behavior data are specifically integrated to form a 13-dimensional three-dimensional employee portrait. Attendance records, performance scores, etc. are used to track dynamic behavior, transforming fragmented human data into a feasible intelligent data pool.

[0026] Step S1 specifically includes: Collect basic employee information, use the employee number as the unique index, and save it in the database. Obtain employee behavior data based on the company's employee system, use the employee number as a common index, and use employee behavior data and employee basic information as employee statistics.

[0027] In this embodiment, the employee number is used as a unique index, and the employee basic information and employee behavior data are connected to form employee statistical data, which improves the structure of the data and prepares data for subsequent employee skill prediction.

[0028] Step S2 specifically includes: The education level and job satisfaction in employee basic information and employee behavior data are coded according to the level of education level and job satisfaction from high to low, and gender, marital status and department are coded using one-hot encoding. Employee basic information and employee behavior data are saved as numerical feature matrices, categorical feature matrices, text feature matrices, and time series feature matrices according to numerical, categorical, textual, and time series types, and all feature matrices are merged to form a multimodal feature set. A deep learning network is constructed using BERT and LSTM network settings. The multimodal feature set is extracted and enhanced through different neural network branches. The numerical features are processed using a fully connected layer activated by RELU. Dropout is used to prevent overfitting for categorical features. Text features are trained stably through LayerNorm. Time series features are activated by ELU to capture nonlinear relationships. The weights of each feature are dynamically calculated and fused through the cross-modal attention mechanism to form a unified feature representation. A multi-layer fully connected network is used to predict and output the skill score.

[0029] In this embodiment, the numerical feature matrix includes historical performance scores and training participation, the categorical feature matrix includes gender, education level, marital status and department, the text feature matrix includes job satisfaction, and the time series feature matrix includes attendance records, team contribution value, years of service and age. By utilizing a deep learning network, through an innovative multimodal feature engineering set and deep learning architecture design, the robustness of the model is improved, and explainable data decision support is provided.

[0030] Step S3 specifically includes: a feature matrix module, an employee stability value prediction module, and an employee position matching module; The feature matrix module includes: calculating attendance abnormality rate based on attendance records of employee behavior data ,in is the attendance abnormality rate of the nth employee. The attendance abnormality rate is the ratio of the number of days with abnormal attendance to the total number of days with attendance. The standard deviation of performance fluctuation for each quarter is calculated based on the historical performance score. , based on training participation, count the number of training sessions employees have taken in the past three months , using attendance abnormality rate, performance fluctuation standard deviation and training times to build employee behavior characteristic matrix , construct the job skill characteristic matrix based on skill score x, team contribution value y and job satisfaction z .

[0031] In this embodiment, part of the employee behavior data is further processed to form data that can specifically express the employee behavior characteristics, and the data is stored separately according to job skill characteristics and employee behavior characteristics to form two feature matrices, which capture the changes in employee capabilities, adapt to various organizational structures, and improve the data-driven effect.

[0032] The employee stability value prediction module includes: using NFM to decompose the employee behavior characteristic matrix to obtain the employee engagement factor and risk propensity factor, the sum of which equals 1; using the logistic regression model to perform model stability training on the employee engagement factor and risk propensity factor to obtain the first weight factor and the second weight factor, where the first weight factor corresponds to the employee engagement factor and the second weight factor corresponds to the risk propensity factor; and calculating the first latent factor. The calculation expression of the first latent factor m is: ; According to the first potential factor, the employee stability value is calculated. The calculation expression of the employee stability value is: ; Among them, p is the employee stability value, e is a mathematical constant, and m is the first potential factor.

[0033] In this embodiment, employee stability is predicted based on the employee behavior matrix. According to the real-time tracking of fluctuations in potential factors such as dedication and risk propensity, real-time monitoring is achieved and employee stability scores are calculated. This effectively realizes the data quantification operation of employee on-the-job stability and provides effective data support for HR's subsequent employee adjustments.

[0034] The employee job matching module includes: using NFM to decompose the job skill feature matrix to obtain the technical ability factor and the management ability factor, the sum of which is equal to 1; using cosine similarity to perform model stability training on the technical ability factor and the management ability factor to obtain the third weight factor and the fourth weight factor, where the third weight factor corresponds to the technical ability factor and the fourth weight factor corresponds to the management ability factor; and calculating the second latent factor. The calculation expression of the second latent factor k is: ; According to the second potential factor, the job matching value is calculated. The calculation expression of the job matching value is: ; Among them, l is the job matching value, e is a mathematical constant, and k is the second potential factor.

[0035] In this embodiment, job matching pairs are predicted based on the employee job skill matrix, and real-time monitoring is achieved based on the fluctuations of potential factors such as technical capabilities and management capabilities tracked in real time. The employee job matching score is calculated, effectively realizing data quantification operations on employee job stability, reducing the costs caused by mismatches, and reducing knowledge loss caused by resignation.

[0036] Step S4 specifically includes: Manually set the distribution histogram of the recommendation score, use the employee stability value as the horizontal distribution value, and the job matching degree as the vertical distribution value, obtain the employee job matching stability coordinates, and map them into the distribution histogram. According to the distribution position of the employee job matching stability coordinates, match them to the corresponding scores to obtain the job recommendation scores. Use K-means density clustering to correspond the job recommendation scores to the employee job matching stability coordinates one by one, and visualize the output to obtain a large screen of employee evaluation density clustering.

[0037] In this embodiment, a two-dimensional feature space is constructed, and the horizontal axis is the employee stability value. , the vertical axis is the job matching value , the distribution on the right is used to simulate employee stability, the distribution on the left is used to simulate job matching, and the score mapping rules are manually set to obtain the job recommendation score. Using K-means density clustering, three cluster centers are obtained: When the stability value is greater than 0.7 and the matching value is greater than 0.8, the employee is a high-potential backbone employee, and the cluster is set to 0; When the stability value is between 0.3 and 0.7 and the matching value is between 0.5 and 0.8, the employees are those who need to be developed, and the cluster is set to 1; When the stability value is less than 0.4 and the matching value is less than 0.5, the employees are high-risk employees and the cluster is set to 2; Set up a visual large screen for employee behavior monitoring, which can query individual employees and output personal development reports.

[0038] The employee evaluation density clustering screen is also used to query the current employee's on-the-job status, job score prediction and on-the-job event trajectory based on the employee number and department. According to the department, it can query the number of employees in the current department, department stability and on-the-job distribution of department personnel.

[0039] In this embodiment, the employee evaluation density clustering screen is further described, and the query function, employee monitoring function and report generation function of the visualization screen are explained, so as to effectively realize the visualization monitoring of human resources.

[0040] Example 2, reference Figure 2 This is another embodiment of the present invention. Different from the first embodiment, this embodiment provides a human resources data classification and statistics system based on big data. In order to verify and illustrate the technical effects adopted in this method, this embodiment adopts traditional technical solutions and the method of the present invention for comparative testing, and compares the test results by means of scientific demonstration to verify the real effect of this method.

[0041] A human resources data classification and statistics system based on big data, including a data collection module, a skill scoring module, a characteristic value prediction module and a visual analysis module; Data collection module, used to collect basic employee information and employee behavior data and save them in the corresponding employee log; The skill scoring module is used to perform label encoding and one-hot encoding on employee statistical data, classify employee basic information and employee behavior data according to their data types, and establish a multimodal feature set for employees. This multimodal feature set is then input into a deep learning model to predict employee skill scores. The characteristic value prediction module is used to extract employee behavior data and employee skill scores, sort the data into an employee behavior characteristic matrix and a job skill characteristic matrix, set the first latent factor based on the behavior characteristic matrix to predict the employee stability value, and set the second latent factor based on the job skill characteristic matrix to predict the job matching value; The visualization analysis module is used to assign job recommendation scores based on employee stability value and job matching degree, and visually output the recommendation score heat density map.

[0042] In this embodiment, based on multi-dimensional data integration and intelligent scoring, the defects of traditional evaluation such as strong subjectivity and single dimension are overcome. Label encoding and one-hot encoding processing ensure the compatibility of heterogeneous data, making the scoring more objective and comprehensive. The characteristic value prediction module innovatively introduces the behavioral characteristic matrix and the job skill matrix, and predicts the employee stability value and job matching value through the first and second latent factors respectively. This two-factor analysis simultaneously quantifies employee loyalty and job adaptability. The visualization module converts complex data into a heat density map, intuitively showing the matching degree between employees and positions, supporting the human resources department to quickly adjust the personnel and job configuration, reducing manual intervention errors, and providing effective digital tools for digital talent management.

[0043] It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a completely hardware embodiment, a completely software embodiment or an embodiment combining software and hardware aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. These computer program instructions may also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which is implemented in the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0044] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A human resources data classification and statistics method based on big data, characterized in that: include: Step S1: Collect employee basic information and employee behavior data and save them as employee statistics; Step S2: performing label encoding and one-hot encoding on the employee statistical data, establishing a multimodal feature set of employees based on the employee basic information and employee behavior data, inputting the multimodal feature set of employees into a deep learning model, and predicting employee skill scores; Step S3: extracting the employee behavior data and employee skill scores, constructing an employee behavior feature matrix and a job skill feature matrix, setting a first latent factor based on the behavior feature matrix to predict the employee stability value, and setting a second latent factor based on the job skill feature matrix to predict the job matching value; Step S4: assigning a job recommendation score based on the employee stability value and the job matching degree, and visually outputting a heat density map of the recommendation score.

2. The human resources data classification and statistics method based on big data according to claim 1, characterized in that: The basic information of employees includes: employee number, age, years of service, education background, gender, marital status and department; The employee behavior data includes: employee number, attendance records, historical performance scores, training participation, job satisfaction and team contribution value.

3. The human resources data classification and statistics method based on big data according to claim 2, characterized in that: Step S1 specifically includes: Collect employee basic information, use the employee number as a unique index, and save it in the database. Obtain employee behavior data based on the company's employee system, use the employee number as a common index, and use the employee behavior data and employee basic information as employee statistics.

4. The human resources data classification and statistics method based on big data according to claim 3, characterized in that: Step S2 specifically includes: The education level and job satisfaction in the employee basic information and employee behavior data are coded according to the level of education level and job satisfaction from high to low, and the gender, marital status and department are coded using a one-hot encoding method. The employee basic information and employee behavior data are saved as a numerical feature matrix, a categorical feature matrix, a text feature matrix, and a time series feature matrix according to the numerical type, categorical type, text type, and time series type, respectively, and all feature matrices are merged to form a multimodal feature set; A deep learning network is constructed using BERT and LSTM network settings. The multimodal feature set is extracted and enhanced through different neural network branches. The numerical features are processed using a fully connected layer activated by RELU. Dropout is used to prevent overfitting of the categorical features. The text features are stably trained through LayerNorm. The temporal features are activated by ELU to capture nonlinear relationships. The weights of each feature are dynamically calculated and fused through a cross-modal attention mechanism to form a unified feature representation. A multi-layer fully connected network is used to predict and output the skill score.

5. The human resources data classification and statistics method based on big data according to claim 4, characterized in that: Step S3 specifically includes: a feature matrix module, an employee stability value prediction module, and an employee position matching module; The feature matrix module includes: calculating the attendance abnormality rate based on the attendance records of the employee behavior data ,in The abnormal attendance rate of the nth employee is the ratio of abnormal attendance days to the total attendance days. The standard deviation of performance fluctuations for each quarter is calculated based on the historical performance score. , based on the training participation rate, count the number of training sessions that employees have taken in the past three months , using the attendance abnormality rate, performance fluctuation standard deviation and training times to construct the employee behavior characteristic matrix , construct a job skill feature matrix based on the skill score x, the team contribution value y and the job satisfaction z .

6. The human resources data classification and statistics method based on big data according to claim 5, characterized in that: The employee stability value prediction module includes: using NFM to decompose the employee behavior feature matrix to obtain an employee engagement factor and a risk propensity factor, wherein the sum of the employee engagement factor and the risk propensity factor is equal to 1; using a logistic regression model to perform model stability training on the employee engagement factor and the risk propensity factor to obtain a first weight factor and a second weight factor, wherein the first weight factor corresponds to the employee engagement factor and the second weight factor corresponds to the risk propensity factor; and calculating a first latent factor. The calculation expression of the first latent factor m is: ； According to the first potential factor, the employee stability value is calculated. The calculation expression of the employee stability value is: ； Among them, p is the employee stability value, e is a mathematical constant, and m is the first potential factor.

7. The human resources data classification and statistics method based on big data according to claim 6, characterized in that: The employee position matching module includes: using NFM to decompose the position skill feature matrix to obtain a technical ability factor and a management ability factor, the sum of which is equal to 1; using cosine similarity to perform model stability training on the technical ability factor and the management ability factor to obtain a third weight factor and a fourth weight factor, wherein the third weight factor corresponds to the technical ability factor and the fourth weight factor corresponds to the management ability factor; calculating a second latent factor, and the calculation expression of the second latent factor k is: ； According to the second potential factor, the position matching value is calculated. The calculation expression of the position matching value is: ； Among them, l is the job matching value, e is a mathematical constant, and k is the second potential factor.

8. The human resources data classification and statistics method based on big data according to claim 7, characterized in that: Step S4 specifically includes: Manually set the distribution histogram of the recommendation score, use the employee stability value as the horizontal distribution value, and the job matching degree as the vertical distribution value, obtain the employee job matching stability coordinates, and map them into the distribution histogram. According to the distribution position of the employee job matching stability coordinates, match the corresponding scores to obtain the job recommendation scores, use K-means density clustering to correspond the job recommendation scores to the employee job matching stability coordinates one by one, and visualize the output to obtain a large screen of employee evaluation density clustering.

9. The human resources data classification and statistics method based on big data according to claim 8, characterized in that: The employee evaluation density clustering screen is also used to query the current employee's on-the-job status, job score prediction and on-the-job event trajectory based on the employee number and department, and to query the number of employees in the current department, department stability and on-the-job distribution of department personnel based on the department.

10. A human resources data classification and statistics system based on big data, which is implemented based on a human resources data classification and statistics method based on big data according to any one of claims 1 to 9, characterized in that: It includes data collection module, skill scoring module, characteristic value prediction module and visual analysis module; The data collection module is used to collect basic employee information and employee behavior data and save them in the corresponding employee log; The skill scoring module is used to perform label encoding and one-hot encoding on the employee statistical data, classify the employee basic information and employee behavior data according to their data types, and establish a multimodal feature set for the employees. The multimodal feature set is input into the deep learning model to predict the employee skill score; The characteristic value prediction module is used to extract the employee behavior data and employee skill scores, sort the data into an employee behavior feature matrix and a job skill feature matrix, set a first latent factor based on the behavior feature matrix to predict the employee stability value, and set a second latent factor based on the job skill feature matrix to predict the job matching value; The visualization analysis module is used to assign job recommendation scores according to the employee stability value and job matching degree, and to visually output a heat density map of the recommendation scores.

Images