Dynamic scheduling optimization system and method based on machine learning

Abstract

The invention provides a dynamic scheduling optimization system and method based on machine learning, and relates to the technical field of enterprise management. According to the invention, by accurately evaluating the employee ability and predicting the work task amount in real time, efficient configuration and flexible scheduling of human resources are realized, the scheduling efficiency and accuracy are significantly improved, the work enthusiasm and satisfaction of employees are enhanced by setting the scheduling level and a reward mechanism, the enterprise operation cost is reduced, and meanwhile, the working efficiency is improved. The introduction of intelligent management improves the management efficiency, enables enterprises to more flexibly cope with the market change, shows strong adaptability and competitiveness, and lays a solid foundation for the sustainable development of the enterprises.

Description

Technical Field

[0001] The present invention relates to the technical field of enterprise management, and in particular, to a dynamic shift scheduling optimization system and method based on machine learning. Background Art

[0002] In enterprise management, the shift scheduling problem has always been a complex and crucial link. Traditional shift scheduling methods often rely on manual judgment and experience-based decision-making, which seems inadequate in the face of a changing market environment and employee needs.

[0003] Traditional shift scheduling methods usually require managers to spend a lot of time on manual calculation and arrangement, which is not only time-consuming and laborious but also prone to errors. Secondly, there is a lack of scientific basis in the shift scheduling process, often relying on the personal experience and intuition of managers, resulting in inaccurate shift scheduling results and difficulty in meeting the actual needs of enterprises.

[0004] Traditional shift scheduling methods often ignore the personal abilities and work efficiencies of employees, resulting in unfair and unreasonable shift scheduling results. During peak periods, employees may face excessive work pressure; while during off-peak periods, employees may be idle, causing waste of human resources. Moreover, due to the unfair and unreasonable shift scheduling results, traditional shift scheduling methods may lead to excessive overtime or long working hours for employees, thus increasing the enterprise's overtime expenses and employee welfare costs.

[0005] Therefore, it is necessary to provide a dynamic shift scheduling optimization system and method based on machine learning to solve the above technical problems. Summary of the Invention

[0006] To solve the above technical problems, the present invention provides a dynamic shift scheduling optimization system and method based on machine learning to solve the problems of low shift scheduling efficiency, unreasonable human resource allocation, and high operating costs.

[0007] A dynamic shift scheduling optimization method provided by the present invention, the optimization method includes the following steps:

[0008] Collect the historical work data and basic information of each employee, and establish a corresponding data set for each employee to obtain the data set corresponding to each employee;

[0009] Based on the data set corresponding to each employee, use statistical methods to evaluate the work ability of each employee to obtain an evaluation result;

[0010] Divide the shift scheduling levels of employees based on historical shift scheduling data, and divide the corresponding reward levels for the shift scheduling levels;

[0011] Obtain the historical work task volume data of the enterprise, and use machine learning technology to train and construct a task volume prediction model based on the historical work task volume data of the enterprise;

[0012] Obtain the current workload of the enterprise in real time according to the set period, and analyze and estimate the workload of the next period through the constructed workload estimation model;

[0013] According to the estimated workload of the next period, divide the emergency scheduling situation of each employee in the current scheduling situation with the scheduling level division and evaluation results of the employees as weights;

[0014] Based on the divided emergency scheduling situation of each employee, make dynamic adjustments based on the current scheduling situation to obtain the updated scheduling table and reward table.

[0015] Preferably, the specific steps for collecting the historical work data and basic information of each employee and establishing a corresponding data set for each employee are as follows:

[0016] Collect the historical work data of each employee from the enterprise's HR system, attendance system or project management system, where the historical work data specifically includes the attendance records, project completion status and work completion degree of the employees;

[0017] Collect the basic information of the employees, specifically including name, age, gender, position and work experience;

[0018] Clean the collected historical work data and basic information of each employee, and organize and establish a corresponding data set for each employee after cleaning.

[0019] Preferably, the specific steps for evaluating the work ability of each employee based on the data set corresponding to each employee and obtaining the evaluation result are as follows:

[0020] After preprocessing the data set corresponding to each employee, use the mean, standard deviation, median or percentile in statistical methods to calculate the average work efficiency of each employee;

[0021] Based on the calculated average work efficiency of each employee, quantify it into corresponding specific evaluation values and output them in the form of a report or table to obtain the evaluation result.

[0022] Preferably, the specific steps for dividing the scheduling levels of employees based on historical scheduling data and dividing the corresponding reward levels are as follows:

[0023] Extract historical scheduling data from the enterprise internal system or database and perform cleaning processing, where the historical scheduling data specifically includes attendance rate, task completion status and additional overtime situation;

[0024] Set indicators based on historical scheduling data, and assign weights to attendance rate, task completion, and overtime respectively according to the set indicators;

[0025] Calculate the comprehensive weight of each employee's schedule by weighted summation, and divide employees into different scheduling levels according to the preset weight level intervals;

[0026] Based on the divided scheduling levels, set corresponding reward levels for each scheduling level.

[0027] Preferably, the specific steps for obtaining the enterprise's historical work task volume data and using machine learning technology to train based on the enterprise's historical work task volume data to construct a task volume prediction model are as follows:

[0028] Collect historical work task volume data from the enterprise system and perform cleaning processing. Among them, the historical work task volume data includes task type, task quantity, and task difficulty;

[0029] Select features related to task volume and task volume information corresponding to the features related to task volume from the historical work task volume data. Among them, the features related to task volume specifically include time, holidays, and project cycle;

[0030] Use the features related to task volume as the input of the training set, and the task volume information corresponding to the features related to task volume as the output. Select linear regression or decision tree in the machine learning model for training to obtain the task volume prediction model.

[0031] Preferably, the specific steps for obtaining the enterprise's current work task volume in real time according to the set cycle and analyzing and predicting the work task volume of the next cycle through the constructed task volume prediction model are as follows:

[0032] Obtain the enterprise's current work task volume data in real time according to the set cycle and perform cleaning and standardization processing;

[0033] Input the enterprise's current work task volume data into the task volume prediction model to analyze and predict the work task volume of the next cycle.

[0034] Preferably, the specific steps for dividing the emergency scheduling situation of each employee in the current scheduling situation according to the predicted work task volume of the next cycle, with the employee's scheduling level division and evaluation results as weights are as follows:

[0035] According to the predicted work task volume of the next cycle, analyze whether the current scheduling situation meets the requirements;

[0036] Set the scheduling weights of different employees according to the employee's scheduling level division and evaluation results;

[0037] According to the preset scheduling weight threshold range, divide the emergency levels of the emergency scheduling of each employee in the current scheduling situation, and set corresponding additional rewards for the emergency levels.

[0038] Preferably, based on the emergency scheduling situations of each employee obtained by division, the specific steps for dynamically adjusting based on the current scheduling situation to obtain an updated scheduling table and reward table are as follows:

[0039] According to the emergency scheduling situations of each employee obtained by division, dynamically add an emergency table to the current scheduling table to obtain an optimized and adjusted scheduling table;

[0040] Based on the obtained optimized and adjusted scheduling table, update the scheduling reward table and additional rewards.

[0041] A dynamic scheduling optimization system based on machine learning, the optimization system includes:

[0042] A data collection module, configured to collect the historical work data and basic information of each employee, and establish a corresponding data set for each employee to obtain a data set corresponding to each employee;

[0043] An evaluation module, configured to evaluate the work ability of each employee based on the data set corresponding to each employee by using statistical methods to obtain an evaluation result;

[0044] A level division module, configured to divide the scheduling levels of employees based on historical scheduling data, and divide the reward levels corresponding to the scheduling levels;

[0045] A model construction module, configured to obtain the historical work task volume data of the enterprise, and use machine learning technology to train and construct a task volume prediction model based on the historical work task volume data of the enterprise;

[0046] An analysis and prediction module, configured to obtain the current work task volume of the enterprise in real time according to a set period, and analyze and predict the work task volume of the next period through the constructed task volume prediction model;

[0047] A scheduling division module, configured to divide the emergency scheduling situations of each employee in the current scheduling situation with the scheduling level division and evaluation result of the employee as weights according to the predicted work task volume of the next period;

[0048] An adjustment module, configured to dynamically adjust based on the emergency scheduling situations of each employee obtained by division, with the current scheduling situation as the basis, to obtain an updated scheduling table and reward table.

[0049] Compared with the related technologies, a dynamic scheduling optimization system and method based on machine learning provided by the present invention has the following beneficial effects:

[0050] Through the precise evaluation of employees' capabilities and real-time prediction of the workload of tasks, the present invention realizes the efficient allocation and flexible scheduling of human resources. It not only significantly improves the scheduling efficiency and accuracy, but also enhances employees' work enthusiasm and satisfaction by setting up scheduling levels and reward mechanisms, reducing the enterprise operation cost. At the same time, the introduction of intelligent management improves the management efficiency, enabling the enterprise to more flexibly respond to market changes, demonstrating strong adaptability and competitiveness, and laying a solid foundation for the sustainable development of the enterprise. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a flowchart of a dynamic scheduling optimization method based on machine learning according to the present invention;

[0052] Figure 2 It is a system block diagram of a dynamic scheduling optimization system based on machine learning according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0053] The present invention will be further described below in conjunction with the drawings and embodiments.

[0054] Embodiment 1

[0055] As Figure 1 shown, a dynamic scheduling optimization method based on machine learning includes the following steps:

[0056] S1. Collect the historical work data and basic information of each employee, and establish a corresponding data set for each employee to obtain the data sets corresponding to each employee;

[0057] S2. Based on the data sets corresponding to each employee, use statistical methods to evaluate the work capabilities of each employee to obtain evaluation results;

[0058] S3. Divide the scheduling levels of employees based on the historical scheduling data, and divide the corresponding reward levels for the scheduling levels;

[0059] S4. Obtain the historical workload data of the enterprise, and use machine learning technology to train and construct a workload prediction model based on the historical workload data of the enterprise;

[0060] S5. Real-time obtain the current workload of the enterprise at a set period, and analyze and predict the workload of the next period through the constructed workload prediction model;

[0061] S6. According to the predicted workload of the next period, divide the emergency scheduling situations of each employee in the current scheduling situation with the scheduling level division and evaluation results of the employees as weights;

[0062] S7. Based on the emergency shift schedules of each employee obtained from the partitioning, make dynamic adjustments based on the current shift schedule to obtain the updated shift schedule and reward table.

[0063] In the specific implementation process, the specific steps of step S1 are as follows:

[0064] S101. Collect the historical work data of each employee from the enterprise's HR system, attendance system, or project management system, where the historical work data specifically includes the attendance records, project completion status, and work completion degree of the employees.

[0065] Specifically, extract the statistical data of the employees' attendance records, project completion status, and work efficiency from the enterprise's HR system, attendance system, or project management system through database queries or API interfaces.

[0066] Exemplarily, in the HR system, obtain the employment time of an employee, whose position is a development engineer; in the attendance system, obtain the attendance record of an employee showing that he was late 2 times this month, but there are no early leave or leave records; in the project management system, an employee participated in 3 projects, and 2 of them have been completed.

[0067] S102. Collect the basic information of the employees, specifically including name, age, gender, position, and work experience.

[0068] Specifically, collect the basic information of the employees, such as name, age, gender, position, and work experience, through employee files or the HR system, etc.

[0069] S103. Clean the historical work data of each employee and the basic information of the employees collected, and after cleaning, organize and establish a dataset corresponding to each employee.

[0070] Specifically, use data processing tools or programming languages (such as Python, R, etc.) to clean and organize the historical work data and basic information of the employees collected to remove duplicate, incorrect, or invalid data, ensure the accuracy and consistency of the data, and organize the data into a dataset corresponding to each employee.

[0071] In the specific implementation process, the specific steps of step S2 are as follows:

[0072] S201. After preprocessing the dataset corresponding to each employee, use the mean, standard deviation, median, or percentile in statistical methods to calculate the average work efficiency of each employee;

[0073] Specifically, according to the evaluation requirements, the datasets corresponding to each employee can be divided and classified according to dimensions such as time, project, or task, so as to calculate work efficiency more accurately; then, indicators such as mean, standard deviation, median, or percentile in statistical methods are used to calculate the work efficiency of employees. In this embodiment, the mean is selected as the indicator, and the average number of tasks completed by each employee per day is calculated.

[0074] S202. Based on the calculated average work efficiency of each employee, quantify it into corresponding specific evaluation values, and output them in the form of a report or table to obtain the evaluation results.

[0075] Specifically, set the evaluation criteria and quantification rules, calculate the average work efficiency of each employee obtained, and convert it according to the evaluation criteria and quantification rules to obtain specific evaluation values. Then, when outputting the evaluation results, data processing tools (such as Excel, Word, etc.) or report generation software (such as Power BI, Tableau, etc.) can be used for collation and presentation. In this embodiment, the evaluation criteria and quantification rules are specifically evaluation levels such as excellent, good, average, and poor. Specifically, an average of ≥5 tasks completed per day is considered excellent, 4 - 5 tasks is considered good, 3 - 4 tasks is considered average, and ≤3 tasks is considered poor. Exemplarily, if the evaluation value of employee A is set as excellent (5.0), the evaluation value of employee B is set as good (4.5), and so on. Finally, the quantified evaluation values are collated and presented in the form of a table, and the table includes information such as the employee's name, position, average work efficiency, evaluation value, and evaluation level.

[0076] In the specific implementation process, the specific steps of step S3 are as follows:

[0077] S301. Extract historical shift scheduling data from the enterprise internal system or database and perform cleaning processing, where the historical shift scheduling data specifically includes attendance rate, task completion situation, and additional overtime situation.

[0078] Specifically, determine the range and time period of the historical shift scheduling data to be extracted, obtain the historical shift scheduling data according to the determined data range and time period, and then perform cleaning processing on it, including removing invalid records, merging duplicate records, and unifying data formats, etc.

[0079] S302. Set indicators based on the historical shift scheduling data, and assign weights to the attendance rate, task completion situation, and additional overtime situation respectively according to the set indicators.

[0080] Specifically, metrics are set for attendance rate, task completion, and overtime hours respectively. Based on the enterprise's requirements, the importance and relevance of the metrics are set manually, and a weight value is assigned to each metric. Specifically, statistical analysis software (such as SPSS, Excel, etc.) can be used for the calculation and verification of weight assignment.

[0081] In this embodiment, according to the enterprise's requirements, attendance rate is crucial for scheduling stability, so a relatively high weight (such as 40%) is assigned to it; task completion efficiency is equally important for business continuity, and a weight of 30% is assigned; while the overtime hours can reflect the dedication of employees, but excessive overtime may have a negative impact on employees' health, so a lower weight (such as 30%) is assigned.

[0082] S303. Calculate the comprehensive weight of each employee's schedule by weighted summation, and divide the employees into different scheduling levels according to the preset weight level range.

[0083] Specifically, a programming language (such as Python, Java, etc.) or statistical analysis software is used to calculate the comprehensive weight value of each employee. Specifically, the score of each metric is multiplied by the corresponding weight, and then the weighted scores of each metric are added together to obtain the total score, which is the comprehensive weight value. Then, according to the preset weight level range, the employees are divided into different scheduling levels. In this embodiment, the set weight level range is: above 80 points is high level, 60 - 80 points is medium level, and below 60 points is low level. Exemplarily, an employee's attendance rate score is 80 points, the task completion efficiency score is 90 points, and the overtime hours score is 70 points. According to the set weights (attendance rate 40%, task completion efficiency 30%, overtime hours 30%), the comprehensive weight value of this employee can be calculated as (80×40% + 90×30% + 70×30%) = 80 points, and then it is divided into the high-level weight level range.

[0084] S304. Set corresponding reward levels for each scheduling level based on the divided scheduling levels.

[0085] Specifically, corresponding reward levels are set for each scheduling level based on the divided scheduling levels. Exemplarily, for employees in the high-level scheduling level, high bonus standards and more promotion opportunities are set as rewards; for employees in the medium-level scheduling level, a certain amount of bonus and training opportunities can be provided; for employees in the low-level scheduling level, improvement suggestions and additional coaching support can be provided.

[0086] In the specific implementation process, the specific steps of step S4 are:

[0087] S401. Collect historical workload data from the enterprise system and perform cleaning processing. Among them, the historical workload data includes task type, task quantity, and task difficulty.

[0088] Specifically, export historical workload data from databases such as the task management system, project management system, or human resource management system of the enterprise system, specifically including task type, task quantity, and task difficulty, and perform cleaning processing on it.

[0089] S402. Select features related to the task volume and the task volume information corresponding to the features related to the task volume from the historical workload data. Among them, the features related to the task volume specifically include time, holidays, and project cycle.

[0090] Specifically, according to the historical workload data, analyze which features are related to the task volume. Among them, the features related to the task volume include time (such as month, quarter, year, etc.), holidays (such as whether it is a holiday, holiday type, etc.), and project cycle (such as project start time, end time, project stage, etc.). Then, evaluate the correlation between these features and the task volume through statistical analysis or machine learning algorithms (such as chi-square test, information gain, etc.). According to the preset correlation threshold, select the features that exceed the correlation threshold. Then, for the selected features, extract their corresponding task volume information.

[0091] S403. Use the features related to the task volume as the input of the training set and the task volume information corresponding to the features related to the task volume as the output. Select linear regression or decision tree in the machine learning model for training to obtain a task volume prediction model.

[0092] Specifically, organize the selected features and their corresponding task volume information into a training data set. Then select a linear regression model or a decision tree model for training, and input the training data set into the selected machine learning model for model training. During the training process, the model will learn the mapping relationship between the input features and the output variables, and adjust the model parameters to minimize the prediction error.

[0093] In the specific implementation process, the specific steps of step S5 are as follows:

[0094] S501. Real-time obtain the current workload data of the enterprise according to the set cycle and perform cleaning and standardization processing.

[0095] Specifically, according to the business requirements of the enterprise, a data acquisition cycle is set. The data acquisition cycle is daily, weekly or monthly. Within the set cycle, the current workload data of the enterprise is obtained in real time through the internal task management system, project management software or other relevant data sources of the enterprise. Then, the obtained current workload data of the enterprise is cleaned to remove duplicate, invalid or abnormal data.

[0096] S502. Input the current workload data of the enterprise into the workload prediction model to analyze and predict the workload for the next cycle.

[0097] In the specific implementation process, the specific steps of step S6 are as follows:

[0098] S601. According to the predicted workload for the next cycle, analyze whether the current work schedule meets the requirements.

[0099] Specifically, compare the predicted workload for the next cycle with the current work schedule to evaluate whether the current work schedule can meet the work requirements for the next cycle. Exemplarily, if an enterprise predicts that the task quantity will increase by 20% in the next week, and the proportion of tasks with high technical requirements in the task type will also increase. When analyzing the current work schedule, it is found that the staffing of some positions with high technical requirements is insufficient, and the working hours of some employees are too compact to cope with the increased future workload.

[0100] When analyzing the current work schedule, it is found that the staffing of some positions with high technical requirements is insufficient, and the working hours of some employees are too compact to cope with the increased future workload.

[0101] S602. Set the scheduling weights for different employees according to the scheduling level classification and evaluation results of the employees.

[0102] Specifically, set the corresponding scheduling weights according to the scheduling levels of the employees, specifically excellent, good, average, and poor levels. It should be noted that employees with higher weights will be given priority in scheduling to ensure that key tasks and tasks with high technical requirements can obtain sufficient personnel support.

[0103] S603. According to the preset scheduling weight threshold range, divide the emergency levels of the emergency scheduling of each employee in the current work schedule, and set corresponding additional rewards for the emergency levels.

[0104] Specifically, different scheduling weight threshold ranges are set. Specifically, the weights are divided into three threshold ranges: high, medium, and low, and corresponding threshold ranges are set. Then, according to the scheduling weights of each employee in the current scheduling situation, they are divided into different emergency levels. The higher the weight of an employee, the higher the emergency level, which means they will be given priority when dealing with emergencies or urgent tasks. Next, corresponding additional rewards are set for employees with different emergency levels.

[0105] In the specific implementation process, the specific steps of step S7 are as follows:

[0106] S701. According to the emergency scheduling situations of each employee divided, dynamically add an emergency schedule to the current schedule to obtain an optimized and adjusted schedule.

[0107] Specifically, dynamically add an emergency schedule to the current schedule. In this embodiment, additional emergency shifts are added to the schedule. It should be noted that the added emergency schedule clearly marks information such as the type, time, location of the emergency shift, and the skills and quantity of employees required. Integrate the dynamically added emergency schedule with the original schedule to form an optimized and adjusted schedule.

[0108] S702. Based on the obtained optimized and adjusted schedule, update the scheduling reward table and additional rewards.

[0109] Embodiment 2

[0110] As Figure 2 shown, a dynamic scheduling optimization system based on machine learning for an application of a dynamic scheduling optimization method based on machine learning specifically includes:

[0111] A data collection module, used to collect the historical work data and basic information of each employee, and establish a corresponding data set for each employee to obtain the data set corresponding to each employee;

[0112] An evaluation module, used to evaluate the work ability of each employee based on the data set corresponding to each employee by using statistical methods to obtain an evaluation result;

[0113] A level division module, used to divide the scheduling levels of employees based on historical scheduling data and divide the corresponding reward levels for the scheduling levels;

[0114] A model construction module, used to obtain the historical work task volume data of the enterprise, and use machine learning technology to train and construct a task volume prediction model based on the historical work task volume data of the enterprise;

[0115] An analysis and prediction module, used to obtain the current work task volume of the enterprise in real time according to a set period, and analyze and predict the work task volume of the next period through the constructed task volume prediction model;

[0116] The shift scheduling division module is used to divide the emergency shift scheduling of each employee in the current shift scheduling according to the estimated workload of the next cycle, with the shift scheduling level division and evaluation results of the employees as weights.

[0117] The adjustment module is used to perform dynamic adjustment based on the emergency shift scheduling of each employee divided, with the current shift scheduling as the basis, to obtain the updated shift schedule and reward table.

[0118] This application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in one process Figure 1 one process or multiple processes and / or blocks Figure 1 or multiple blocks.

[0119] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program, and this program can be stored in a computer-readable storage medium. The storage medium includes read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc memories, magnetic disk memories, tape memories, or any other medium that can be used to carry or store data and is computer-readable.

[0120] It should also be noted that the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, commodity or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, commodity or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, commodity or device comprising the element.

Claims

1. A dynamic shift scheduling optimization method based on machine learning, characterized in that The optimization method includes the following steps: Collect the historical work data and basic information of each employee, and establish a corresponding dataset for each employee to obtain the dataset corresponding to each employee; Based on the datasets corresponding to each employee, use statistical methods to evaluate the work ability of each employee to obtain an evaluation result; Based on the historical shift scheduling data, divide the shift scheduling levels of employees and divide the corresponding reward levels for the shift scheduling levels; Obtain the historical work task volume data of the enterprise, and use machine learning technology to train based on the historical work task volume data of the enterprise to construct a task volume prediction model; Obtain the current work task volume of the enterprise in real time according to the set period, and analyze and predict the work task volume of the next period through the constructed task volume prediction model; According to the predicted work task volume of the next period, use the shift scheduling level division and evaluation result of employees as weights to divide the emergency shift scheduling situations of each employee in the current shift scheduling situation; Based on the divided emergency shift scheduling situations of each employee, make dynamic adjustments based on the current shift scheduling situation to obtain an updated shift schedule and reward table.

2. The dynamic shift scheduling optimization method based on machine learning according to claim 1, wherein, The specific steps of collecting the historical work data and basic information of each employee, and establishing a corresponding dataset for each employee to obtain the dataset corresponding to each employee are as follows: Collect the historical work data of each employee from the enterprise's HR system, attendance system or project management system, where the historical work data specifically includes the attendance records, project completion status and work completion degree of employees; Collect the basic information of employees, specifically including name, age, gender, position and work experience; Clean the historical work data and basic information of each employee collected, and organize and establish a dataset corresponding to each employee after cleaning.

3. A dynamic shift scheduling optimization method based on machine learning according to claim 1, characterized in that The specific steps of using statistical methods to evaluate the work ability of each employee based on the datasets corresponding to each employee to obtain an evaluation result are as follows: After preprocessing the datasets corresponding to each employee, use the mean, standard deviation, median or percentile in statistical methods to calculate the average work efficiency of each employee; Based on the calculated average work efficiency of each employee, quantify it into corresponding specific evaluation values and output them in the form of a report or table to obtain an evaluation result.

4. A dynamic shift scheduling optimization method based on machine learning according to claim 1, characterized in that The specific steps of dividing the shift scheduling levels of employees based on the historical shift scheduling data and dividing the corresponding reward levels for the shift scheduling levels are as follows: Extract the historical shift scheduling data from the enterprise internal system or database and perform cleaning processing, where the historical shift scheduling data specifically includes attendance rate, task completion status and additional overtime situation; Set indicators based on the historical shift scheduling data, and assign weights to the attendance rate, task completion status and additional overtime situation according to the set indicators; Calculate the comprehensive weight of each employee's shift scheduling by weighted summation, and divide the employees into different shift scheduling levels according to the preset weight level interval; Based on the divided shift scheduling levels, set corresponding reward levels for each shift scheduling level.

5. A dynamic shift scheduling optimization method based on machine learning according to claim 1, characterized in that The specific steps of obtaining the historical work task volume data of the enterprise and using machine learning technology to train based on the historical work task volume data of the enterprise to construct a task volume prediction model are as follows: Collect historical workload data from the enterprise system and perform cleaning processing. Among them, the historical workload data includes task type, task quantity, and task difficulty. Select features related to the workload and the corresponding workload information of the features related to the workload from the historical workload data. Among them, the features related to the workload specifically include time, holidays, and project cycle. Use the features related to the workload as the input of the training set, and the corresponding workload information of the features related to the workload as the output. Select linear regression or decision tree in the machine learning model for training to obtain a workload prediction model.

6. The dynamic shift scheduling optimization method based on machine learning according to claim 1, wherein The specific steps for obtaining the current workload of the enterprise in real time according to the set cycle and analyzing and predicting the workload of the next cycle through the constructed workload prediction model are as follows: Obtain the current workload data of the enterprise in real time according to the set cycle and perform cleaning and standardization processing. Input the current workload data of the enterprise into the workload prediction model to analyze and predict the workload of the next cycle.

7. A dynamic shift scheduling optimization method based on machine learning according to claim 1, characterized in that The specific steps for dividing the emergency scheduling situation of each employee in the current scheduling situation according to the predicted workload of the next cycle, with the scheduling level division and evaluation results of employees as weights, are as follows: Analyze whether the current scheduling situation meets the requirements according to the predicted workload of the next cycle. Set the scheduling weights of different employees according to the scheduling level division and evaluation results of employees. According to the preset scheduling weight threshold range, divide the emergency levels of the emergency scheduling of each employee in the current scheduling situation, and set corresponding additional rewards for the emergency levels.

8. A dynamic shift scheduling optimization method based on machine learning according to claim 1, characterized in that The specific steps for dynamically adjusting based on the divided emergency scheduling situations of each employee and obtaining the updated scheduling table and reward table with the current scheduling situation as the basis are as follows: According to the divided emergency scheduling situations of each employee, dynamically add an emergency table to the current scheduling table to obtain an optimized and adjusted scheduling table. Based on the obtained optimized and adjusted scheduling table, update the scheduling reward table and additional rewards.

9. A dynamic shift scheduling optimization system based on machine learning, which is applied to a dynamic shift scheduling optimization method based on machine learning according to any one of claims 1-8, characterized in that, The optimization system includes: A data collection module for collecting the historical work data and basic information of each employee, and establishing a corresponding data set for each employee to obtain the data set corresponding to each employee. An evaluation module for evaluating the work ability of each employee using statistical methods based on the data set corresponding to each employee to obtain an evaluation result. A level division module for dividing the scheduling levels of employees based on historical scheduling data and dividing the corresponding reward levels for the scheduling levels. A model construction module for obtaining the historical workload data of the enterprise and using machine learning technology to train and construct a workload prediction model based on the historical workload data of the enterprise. An analysis and prediction module for obtaining the current workload of the enterprise in real time according to the set cycle and analyzing and predicting the workload of the next cycle through the constructed workload prediction model. A scheduling division module for dividing the emergency scheduling situations of each employee in the current scheduling situation according to the predicted workload of the next cycle, with the scheduling level division and evaluation results of employees as weights. Adjustment module, which is used to dynamically adjust based on the emergency shift schedules of each divided employee, taking the current shift schedule as the basis, to obtain the updated shift schedule and reward table.

Images