Change day personnel demand prediction method and device, equipment and storage medium
By combining the Prophet model and the Gaussian Hidden Markov model, the problem of inaccurate personnel demand forecasts in the existing technology is solved, and more accurate personnel demand forecasts are achieved, which reduces risks and costs and improves service quality.
Patent Information
- Application Number
- CN202411686498.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-22
- Publication Date
- 2025-02-28
AI Technical Summary
When predicting personnel needs on the change day, the prediction results are not accurate enough, resulting in improper allocation of human resources and increasing risks and costs.
By combining the Prophet model with the Gaussian Hidden Markov model, the Prophet model is used to extract change trends, seasonal fluctuations and holiday impacts from weekday operation volume timing data. The Gaussian Hidden Markov model is used to model and predict special changes in the operation volume on the change day, and thus accurately predict the personnel needs on the change day.
It realizes a more accurate prediction of personnel needs on the day of change, reduces the risks and costs caused by improper human resources allocation, ensures that business-related maintenance tasks can be completed efficiently and safely on the day of change, and improves service quality.
Smart Images
Figure CN119539188A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method, device, equipment and storage medium for predicting personnel demand on a change date. Background Art
[0002] In the field of operation and maintenance management, the effective scheduling of operation and maintenance personnel is crucial to ensure the efficient and stable operation of the business. Especially on change days (such as specific days for system upgrades, maintenance or other key guarantees), which usually involve complex technical operations and high-risk management, it is particularly important to correctly predict personnel needs.
[0003] At present, the staff demand on the change date is usually predicted manually based on experience judgment and historical data analysis. However, the prediction results of the staff demand on the change date are not accurate enough. Summary of the invention
[0004] The present application provides a method, device, equipment and storage medium for predicting personnel demand on a change date, so as to solve the problem that the prediction results of personnel demand on a change date are not accurate enough when predicting personnel demand on a change date using the current method.
[0005] In a first aspect, the present application provides a method for predicting personnel demand on a change date, comprising:
[0006] In response to a personnel demand quantity forecasting instruction for a target change date, obtaining time series data of the operation volume on weekdays before the target change date and time series data of the operation volume on the change date;
[0007] Input the daily operation volume time series data into the Prophet model to predict the operation volume, and obtain the first predicted operation volume and target residual term output by the Prophet model;
[0008] Input the time series data of the operation volume on the changed day and the target residual term into the Gaussian hidden Markov model to predict the operation volume, and obtain the second predicted operation volume output by the Gaussian hidden Markov model;
[0009] Based on the first predicted operation volume and the second predicted operation volume, obtaining a target operation volume corresponding to the target change date;
[0010] Based on the target operation volume, determine the target number of personnel required corresponding to the target change date.
[0011] In a second aspect, the present application provides a change day personnel demand prediction device, comprising:
[0012] A first acquisition module is used to respond to the personnel demand quantity prediction instruction for the target change date and acquire the time series data of the operation volume on weekdays before the target change date and the time series data of the operation volume on the change date;
[0013] The first prediction module is used to input the daily operation volume time series data into the Prophet model to predict the operation volume, and obtain the first predicted operation volume and target residual term output by the Prophet model;
[0014] The second prediction module is used to input the time series data of the operation volume on the changed day and the target residual term into the Gaussian hidden Markov model to predict the operation volume, and obtain the second predicted operation volume output by the Gaussian hidden Markov model;
[0015] A second acquisition module, configured to obtain a target operation amount corresponding to the target change date based on the first predicted operation amount and the second predicted operation amount;
[0016] The determination module is used to determine the target number of personnel required corresponding to the target change date based on the target operation volume.
[0017] In a third aspect, the present application provides an electronic device, comprising: a processor, and a memory communicatively connected to the processor;
[0018] Memory stores computer-executable instructions;
[0019] The processor executes the computer-executable instructions stored in the memory to implement the above first aspect and / or various possible implementations of the first aspect.
[0020] In a fourth aspect, the present application provides a computer-readable storage medium, in which computer program instructions are stored. When the computer program instructions are executed, the above first aspect and / or various possible implementation methods of the first aspect are implemented.
[0021] In a fifth aspect, the present application provides a computer program product, including a computer program, which implements the above first aspect and / or various possible implementation methods of the first aspect when the computer program is executed.
[0022] The change date personnel demand prediction method, device, equipment and storage medium provided in the present application, after obtaining the time series data of the daily operation volume before the target change date and the time series data of the change date operation volume, input the time series data of the daily operation volume into the Prophet model for operation volume prediction, and obtain the first predicted operation volume and target residual term output by the Prophet model; input the time series data of the change date operation volume and the target residual term into the Gaussian hidden Markov model for operation volume prediction, and obtain the second predicted operation volume output by the Gaussian hidden Markov model; based on the first predicted operation volume and the second predicted operation volume, obtain the target operation volume corresponding to the target change date, realize the combination of the Prophet model and the Gaussian hidden Markov model, and at the same time mine the implicit state and significant time regularity of the time series data to obtain the target operation volume corresponding to the target change date, so that the target personnel demand quantity corresponding to the target change date can be determined according to the target operation volume, and the target personnel demand quantity can be determined more accurately, reducing the risks and costs caused by improper allocation of human resources, ensuring that the maintenance tasks corresponding to the business can be completed efficiently and safely on the target change date, and improving the service quality. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.
[0024] Figure 1 A schematic diagram of an application scenario provided for an embodiment of the present application;
[0025] Figure 2 Schematic diagram of the process of the change day personnel demand prediction method provided in the embodiment of the present application Figure 1 ;
[0026] Figure 3 Schematic diagram of the process of the change day personnel demand prediction method provided in the embodiment of the present application Figure 2 ;
[0027] Figure 4 A flowchart of a training method for a Prophet model provided in one embodiment of the present application;
[0028] Figure 5 A schematic diagram of a flow chart of a training method for a Gaussian Hidden Markov Model provided in one embodiment of the present application;
[0029] Figure 6 A schematic diagram of the structure of a device for predicting personnel demand on a change date provided in one embodiment of the present application;
[0030] Figure 7 A schematic diagram of the structure of an electronic device provided in one embodiment of the present application.
[0031] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION
[0032] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.
[0033] First, the terms involved in this application are explained:
[0034] Deep Learning (DL) is a new research direction in the field of Machine Learning (ML). Deep learning is introduced into machine learning to make it closer to its original goal - Artificial Intelligence (AI).
[0035] A time series prediction algorithm (Prophet algorithm) is an open source time series prediction algorithm that combines time series decomposition and machine learning algorithms, and can predict time series data (i.e. time series data) with missing data and abnormal data; it is suitable for various prediction problems with potential special characteristics, including a wide range of business time series problems, and has a better fitting effect on the detection of time series trend change points, seasonality, holidays and emergencies.
[0036] Markov Chain (MC) is a concept in probability theory and mathematical statistics, which describes a random process with a "memoryless" property. In a Markov chain, given the current state of the system, the probability distribution of the next state depends only on the previous state and has nothing to do with earlier states. A Markov chain can be either discrete or continuous.
[0037] The Hidden Markov Model (HMM) is a probabilistic model of time series, which describes the process of randomly generating a random sequence of unobservable states by a hidden Markov chain, and then generating a random sequence of observations by generating an observation from each state. The Hidden Markov Model is based on two basic assumptions: one is that the hidden state at any moment is only related to the hidden state at the previous moment, and the other is that the observation at any moment is only related to the hidden state at the current moment.
[0038] Gaussian Hidden Markov Model (GaussianHMM) is a statistical model used to analyze time series data (time series data) with hidden Markov characteristics; Gaussian Hidden Markov Model is a special hidden Markov model whose observation state follows a continuous Gaussian distribution and is often used in HMM models with continuous observation states.
[0039] Long Short-Term Memory (LSTM) is a time recurrent neural network that is used to solve the long-term dependency problem of general recurrent neural networks (RNNs). It is suitable for processing and predicting important events with relatively long intervals and delays in time series. All RNNs have a chain form of repeated neural network modules. In standard RNNs, the repeated structural module has only a very simple structure, for example, a tanh layer (a key component in RNNs).
[0040] Convolutional Neural Network (CNN) is a feed-forward neural network, usually composed of one or more convolutional layers and fully connected layers (corresponding to the classic neural network NN). In addition, it can also include a pooling layer. The structure of CNN is relatively simple and can be trained using the back-propagation algorithm.
[0041] In the field of operation and maintenance management, the effective scheduling of operation and maintenance personnel is crucial to ensure the efficient and stable operation of the business. Especially on change days (such as specific days for system upgrades, maintenance or other key guarantees), which usually involve complex technical operations and high-risk management, it is particularly important to correctly predict personnel needs.
[0042] At present, manual methods are usually used to predict the personnel needs on the change day based on experience judgment and historical data analysis. However, predicting the personnel needs on the change day through the above method is time-consuming and the prediction results are not accurate enough. In operation and maintenance guarantee, due to the lack of systematic analysis of the operation volume on the change day and scientific evaluation of personnel needs, work planning and manpower allocation are carried out in a subjective judgment manner. There is a lack of a standardized, reasonable and unified method, which makes it difficult for scheduling personnel to reflect the professionalism of their work and unable to adapt to the rapid changes and increasing complexity of the business environment.
[0043] In addition, with the development of artificial intelligence technology, especially the excellent performance of deep learning neural networks in prediction and classification tasks, it is possible to use artificial intelligence technology to predict the manpower demand on the change day. Neural network models can process a large amount of complex data and learn patterns from them, which is more suitable for dynamic and changing business environments. In related technologies, for example, personnel demand can be predicted based on the back propagation (BP) neural network model, workload can be predicted based on the LSTM deep learning model with dual attention mechanism, and the probability of system failure can be predicted based on the Prophet model. However, the existing human resource prediction technology is often based on conventional working conditions and only considers historical business volume data, without considering emergencies as factors affecting workload, and cannot be directly applied in complex working environments such as change days. Therefore, in view of the personnel demand on the change day, an efficient and high-precision solution for predicting the personnel demand on the change day is urgently needed.
[0044] Based on the above problems, the present application provides a method, device, equipment and storage medium for predicting personnel demand on the change day. By combining the Prophet model and the Gaussian hidden Markov model, the Prophet model is used as the main framework to extract the change trend, seasonal fluctuations and holiday impact components from the time series data of the daily operation volume and make predictions; the Gaussian hidden Markov model is used to introduce the changes in the operation volume during special change days, and this is used to model and predict the special event impact information hidden in the residual extracted by the Prophet model, ultimately achieving accurate prediction of the number of personnel required on the change day (such as operation and maintenance personnel), reducing the risks and costs caused by improper human resource allocation, ensuring that the maintenance tasks corresponding to the business can be completed efficiently and safely on the change day, and improving service quality.
[0045] In the following, the application scenarios of the solutions provided in this application are first illustrated.
[0046] Figure 1 This is a schematic diagram of an application scenario provided by an embodiment of the present application. Figure 1 As shown, in this application scenario, the administrator triggers a personnel demand quantity prediction instruction for the target change date through the client 101; the client 101 sends the instruction to the server 102, and the server 102 determines the target personnel demand quantity corresponding to the target change date according to the change date personnel demand prediction method provided in the embodiment of the present application, and sends the target personnel demand quantity to the client 101; the client 101 displays the target personnel demand quantity for the administrator to facilitate the administrator to allocate personnel according to the target personnel demand quantity.
[0047] It should be noted that Figure 1 This is only a schematic diagram of an application scenario provided by the embodiment of the present application. Figure 1 does not limit the equipment included in Figure 1 The positional relationship between the devices is limited.
[0048] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.
[0049] Figure 2 Schematic diagram of the process of the change day personnel demand prediction method provided in the embodiment of the present application Figure 1 The method of the embodiment of the present application can be applied to an electronic device, which can be a server or a server cluster. Figure 2 As shown, the method of the embodiment of the present application includes:
[0050] S201 . In response to a personnel demand quantity prediction instruction for a target change date, obtain the time series data of the operation volume on weekdays before the target change date and the time series data of the operation volume on the change date.
[0051] In the embodiment of the present application, the personnel demand quantity prediction instruction for the target change day can be input by the user to the electronic device executing the embodiment of the present method, or sent by other devices to the electronic device executing the embodiment of the present method. In this step, the change day is, for example, a specific day for system upgrade, maintenance or other key guarantees, and the target change day is, for example, Thursday or Saturday; the operation volume can be understood as the total number of operation steps for the target service, and the personnel are the operators who perform the operation steps. Exemplarily, assuming that the current time is Wednesday, taking the target change day as Thursday as an example, in response to the personnel demand quantity prediction instruction for Thursday, the weekday operation volume time series data and the change day operation volume time series data before Thursday are obtained. Among them, the weekday operation volume time series data can include the timestamp corresponding to each day of the weekday and the operation volume corresponding to the timestamp, and the weekday operation volume time series data does not include the change day operation volume time series data. The change day operation volume time series data can include the change day operation volume (such as the change day operation volume on Thursday or the change day operation volume on Saturday) and whether there is an emergency event. After obtaining the time series data of the daily operation volume and the time series data of the change day operation volume, the time series data of the daily operation volume and the time series data of the change day operation volume can be preprocessed respectively to obtain the preprocessed time series data of the daily operation volume and the preprocessed time series data of the change day operation volume. Among them, the preprocessing may include, for example, data cleaning, data format conversion processing and data normalization processing. For specific how to preprocess the time series data of the daily operation volume and the time series data of the change day operation volume, respectively, please refer to the subsequent embodiments.
[0052] S202: Input the daily operation volume time series data into the Prophet model to perform operation volume forecasting, and obtain the first forecast operation volume and target residual term output by the Prophet model.
[0053] In this step, the Prophet model is obtained by training based on the training time series data of the daily operation volume. For specific information on how to train and obtain the Prophet model, please refer to the subsequent embodiments. The operation volume prediction of the Prophet model can be implemented by calling the fbprophet library (a library for time series prediction). Exemplarily, after obtaining the time series data of the daily operation volume, the time series data of the daily operation volume can be input into the Prophet model, and the trend of the daily operation volume change, seasonal fluctuations, and holiday impact components can be extracted to predict the operation volume, and the first predicted operation volume and target residual term for the target change day output by the Prophet model can be obtained.
[0054] S203, inputting the time series data of the changed daily operation volume and the target residual term into the Gaussian hidden Markov model to predict the operation volume, and obtaining a second predicted operation volume output by the Gaussian hidden Markov model.
[0055] In this step, the Gaussian hidden Markov model is obtained by training based on the training time series data of the change day operation volume and the training residual items obtained when training the Prophet model. For how to train the Gaussian hidden Markov model, please refer to the subsequent embodiments. The operation volume prediction of the Gaussian hidden Markov model can be implemented by calling the hmmlearn library (a Python library for learning and inferring HMM). Exemplarily, after obtaining the time series data of the change day operation volume and the target residual items, the time series data of the change day operation volume and the target residual items can be input into the Gaussian hidden Markov model for operation volume prediction, and the second predicted operation volume for the target change day output by the Gaussian hidden Markov model can be obtained.
[0056] S204: Obtain a target operation amount corresponding to the target change date based on the first predicted operation amount and the second predicted operation amount.
[0057] For example, after obtaining the first predicted operation amount and the second predicted operation amount, the first predicted operation amount and the second predicted operation amount can be fused to obtain the target operation amount corresponding to the target change day. For specific information on how to perform the fusion process, please refer to the subsequent embodiments. Taking the change days as Thursday and Saturday as an example, the final prediction result of the change day operation amount on Thursday and the final prediction result of the change day operation amount on Saturday can be obtained.
[0058] S205. Determine the target number of personnel required for the target change date based on the target operation volume.
[0059] Exemplarily, the correspondence between the operation volume and the number of personnel required can be preset. Optionally, in order to have sufficient redundant manpower on the change day and ensure production safety on the change day, for example, 120% of the average operation volume of the operators on weekdays can be used as the maximum workload of a single operator. In this step, after obtaining the target operation volume corresponding to the target change day, the target number of personnel required for the target change day can be determined based on the target operation volume and the correspondence, so that personnel can be allocated according to the target number of personnel required for the target change day.
[0060] The change day personnel demand prediction method provided in the embodiment of the present application, after obtaining the time series data of the daily operation volume before the target change day and the time series data of the change day operation volume, inputs the time series data of the daily operation volume into the Prophet model for operation volume prediction, and obtains the first predicted operation volume and the target residual term output by the Prophet model; inputs the time series data of the change day operation volume and the target residual term into the Gaussian hidden Markov model for operation volume prediction, and obtains the second predicted operation volume output by the Gaussian hidden Markov model; based on the first predicted operation volume and the second predicted operation volume, obtains the target operation volume corresponding to the target change day, realizes the combination of the Prophet model and the Gaussian hidden Markov model, and at the same time mines the implicit state and significant time regularity of the time series data to obtain the target operation volume corresponding to the target change day, so that the target personnel demand quantity corresponding to the target change day can be determined according to the target operation volume, and the target personnel demand quantity can be determined more accurately, reducing the risks and costs caused by improper allocation of human resources, ensuring that the maintenance tasks corresponding to the business can be completed efficiently and safely on the target change day, and improving the service quality.
[0061] Figure 3 Schematic diagram of the process of the change day personnel demand prediction method provided in the embodiment of the present application Figure 2 Based on the above embodiments, the present application further illustrates the method for predicting personnel demand on the change date. Figure 3 As shown, the method of the embodiment of the present application may include:
[0062] S301 . In response to a personnel demand quantity prediction instruction for a target change date, obtain the time series data of the operation volume on weekdays before the target change date and the time series data of the operation volume on the change date.
[0063] The detailed description of this step can be found in Figure 2 The relevant description of S201 in the illustrated embodiment will not be repeated here.
[0064] In the embodiment of the present application, Figure 2 The step S202 may further include the following two steps S302 and S303:
[0065] S302: preprocess the daily operation volume time series data to obtain preprocessed daily operation volume time series data, wherein the preprocessing includes at least one of data cleaning, data format conversion processing and data normalization processing, and the data cleaning includes at least one of deleting duplicate data, missing data processing and abnormal data processing.
[0066] Exemplarily, first, the daily operation volume time series data is cleaned, such as removing duplicate data, processing missing data and removing abnormal data. Among them, the missing data and abnormal data are taken as the daily average of the previous month. Secondly, the data format conversion process is performed, and the data format conversion process is to convert the data format in the daily operation volume time series data into the data format required by the model. For example, in order to meet the requirements of the Prophet model input data, the daily operation volume time series data can be divided into two columns and its data format is converted, wherein one column is in date (datetime) format, the column (column) name is represented by ds, and the other column is int64 format, the column name is represented by y. Considering that the data value range is too different and may cause the problem of model weight allocation error, therefore, it is necessary to reduce the absolute value range of the data, such as the log logarithmic transformation (a mathematical processing method) method can be used to normalize the daily operation volume time series data to obtain the pre-processed daily operation volume time series data.
[0067] S303: Input the preprocessed daily operation volume time series data into the Prophet model to perform operation volume prediction, and obtain the first predicted operation volume and target residual term output by the Prophet model.
[0068] In this step, after obtaining the preprocessed weekday operation volume time series data, the preprocessed weekday operation volume time series data can be input into the Prophet model for operation volume prediction to obtain the first predicted operation volume and target residual term output by the Prophet model.
[0069] In the embodiment of the present application, Figure 2 The step S203 may further include the following two steps S304 and S305:
[0070] S304 , preprocessing the time series data of the change day operation volume to obtain the preprocessed time series data of the change day operation volume.
[0071] Exemplarily, referring to the example of step S302, the time series data of the change day operation volume can be preprocessed to obtain the preprocessed time series data of the change day operation volume. Among them, the preprocessing can include data cleaning, data format conversion processing and data normalization processing, and the data cleaning can include duplicate data removal, missing data processing and abnormal data processing. When the time series data of the change day operation volume is cleaned, if the sudden emergency event column of a certain date is displayed as true (True), it means that a sudden emergency event occurred on that day. Accordingly, the operation volume data of that day is marked as abnormal data, and the missing data and abnormal data are both taken as the daily average of the previous month.
[0072] It should be noted that the embodiment of the present application does not limit the order in which steps S304, S302 and S303 are executed.
[0073] S305 , input the pre-processed time series data of the changed daily operation volume and the target residual term into the Gaussian hidden Markov model to predict the operation volume, and obtain a second predicted operation volume output by the Gaussian hidden Markov model.
[0074] In this step, after obtaining the preprocessed time series data of the changed daily operation volume and the target residual term, the preprocessed time series data of the changed daily operation volume and the target residual term can be input into the Gaussian hidden Markov model for operation volume prediction to obtain the second predicted operation volume output by the Gaussian hidden Markov model.
[0075] In the embodiment of the present application, Figure 2 The step S204 may further include the following step S306:
[0076] S306: Perform a weighted sum of the first predicted operation amount and the second predicted operation amount to obtain a target operation amount corresponding to the target change date.
[0077] For example, a first weight corresponding to the first predicted operation amount and a second weight corresponding to the second predicted operation amount may be preset. After obtaining the first predicted operation amount and the second predicted operation amount, a weighted sum may be performed based on the first predicted operation amount, the first weight, the second predicted operation amount and the second weight to obtain the target operation amount corresponding to the target change date.
[0078] S307. Determine the target number of personnel required for the target change date based on the target operation volume.
[0079] The detailed description of this step can be found in Figure 2 The relevant description of S205 in the illustrated embodiment will not be repeated here.
[0080] The change day personnel demand prediction method provided in the embodiment of the present application, after obtaining the weekday operation volume time series data and the change day operation volume time series data before the target change day, pre-processes the weekday operation volume time series data, inputs the pre-processed weekday operation volume time series data into the Prophet model for operation volume prediction, and obtains the first predicted operation volume and the target residual term output by the Prophet model; pre-processes the change day operation volume time series data, inputs the pre-processed change day operation volume time series data and the target residual term into the Gaussian hidden Markov model for operation volume prediction, and obtains the second predicted operation volume output by the Gaussian hidden Markov model; performs weighted summation of the first predicted operation volume and the second predicted operation volume to obtain the target operation volume corresponding to the target change day, realizes the combination of the Prophet model and the Gaussian hidden Markov model to obtain the target operation volume corresponding to the target change day, so that the target personnel demand quantity corresponding to the target change day can be determined according to the target operation volume, and the target personnel demand quantity can be determined more accurately, reducing the risks and costs caused by improper human resource allocation, ensuring that the maintenance tasks corresponding to the business can be completed efficiently and safely on the target change day, and improving the service quality.
[0081] Based on the above embodiments, Figure 4 A flow chart of a training method for a Prophet model provided in one embodiment of the present application. Figure 4 As shown, the method of the embodiment of the present application may include:
[0082] S401. Acquire daily operation volume training time series data, where the daily operation volume training time series data includes a timestamp and a reference operation volume corresponding to the timestamp.
[0083] In this step, the daily operation amount training time series data can be obtained based on the historical operation amount time series data, and the daily operation amount training time series data includes a timestamp and a reference operation amount (i.e., the actual operation amount) corresponding to the timestamp. After obtaining the daily operation amount training time series data, in order to facilitate the Prophet model to process the daily operation amount training time series data, the daily operation amount training time series data can be preprocessed with reference to the example of step S302 to obtain the preprocessed daily operation amount training time series data, and the preprocessed daily operation amount training time series data is used for the training of the Prophet model. Among them, the preprocessing can include, for example, data cleaning, data format conversion processing and data normalization processing, and the data cleaning can include, for example, removing duplicate data, missing data processing and abnormal data processing.
[0084] Optionally, based on the historical operation volume time series data, daily operation volume test time series data can be obtained to evaluate the performance of the Prophet model finally trained.
[0085] Exemplarily, a daily operation volume training data set may be obtained based on the historical operation volume time series data, and the daily operation volume training data set may be decomposed into daily operation volume training time series data and daily operation volume test time series data.
[0086] S402: Input the daily operation volume training time series data into the Prophet model for fitting, and obtain the training prediction operation volume and training residual term corresponding to the timestamp output by the Prophet model.
[0087] Exemplarily, the operation volume prediction of the Prophet model can be realized by calling the fbprophet library. First, the Prophet model is constructed as: y(t) = g(t) + s(t) + h(t) + ε(t); wherein g(t) represents the variation function of the non-periodic trend obtained by decomposing the daily operation volume training time series data, which can also be called the trend term; s(t) represents the variation function of the periodic trend obtained by modeling the periodic changes of the daily operation volume training time series data, such as daily, weekly or annual seasonal changes, s(t) can also be called the periodic term, and the modeling method of this term is the same as that of the exponential smoothing method; h(t) represents the random influence of holidays on the daily operation volume training time series data, which can also be called the holiday term, and h(t) reflects the irregular changes of holidays that last from one day to several days; ε(t) represents the residual term, which is used to represent the fluctuations not predicted in the model.
[0088] By inputting the training time series data of the daily operation volume into the Prophet model and fitting the Prophet model, the training predicted operation volume and training residual item corresponding to the timestamp output by the Prophet model can be obtained. Specifically, the training time series data of the daily operation volume is input into the Prophet model, and the training time series data of the daily operation volume is decomposed into trend items, cycle items, holiday items and residual items through the Prophet model, and the training predicted operation volume of the trend item, cycle item and holiday item is generated according to the timestamp in the training time series data of the daily operation volume. Among them, when obtaining the trend item, the parameter growth of the Prophet model can be set to 'logistic', that is, the trend item g(t) is fitted through the logistic regression growth model (corresponding to Formula 1):
[0089]
[0090] Where C(t) represents the maximum load of daily operations at timestamp t; k represents the growth rate; m represents the offset parameter; a(t) T δ represents the change in growth rate; t represents the timestamp; exp() represents the exponential function with the natural constant e as the base; a(t) T γ represents the change value of the offset.
[0091] When obtaining the periodic term, the periodic term s(t) can be fitted by a finite-order Fourier series (corresponding to Formula 2):
[0092]
[0093] Where t represents the timestamp; a n and b n represents the Fourier coefficient; N represents the order of the Fourier series corresponding to the periodic term; P represents the period of the operation quantity time series data.
[0094] Get the holiday items through the following formula 3:
[0095]
[0096] in, κ=(κ1,...,κ L ) T ; t represents the timestamp, D represents the impact period of holidays, κ represents the parameter vector, and L represents the order of the Fourier series corresponding to the holiday term.
[0097] Then, in the process of instantiating the Prophet model object, since the data changes will be affected by seasons, weeks and days, and there is a certain regularity, these three parameters can be set to True. In addition, the data will also be affected by holidays, so the parameter date setting of holidays is required.
[0098] Finally, the data is fitted through the fit() method of the Prophet model, and the training prediction operation volume and training residual term corresponding to the timestamp are output in the forecast table of the Prophet model.
[0099] S403: Based on the reference operation amount corresponding to the timestamp and the training prediction operation amount, adjust the parameters of the Prophet model to obtain a trained Prophet model.
[0100] In this step, after the training prediction operation amount corresponding to the timestamp is obtained, the training prediction operation amount corresponding to the timestamp can be compared with the reference operation amount, and the parameters of the Prophet model can be adjusted to obtain the trained Prophet model.
[0101] The training method of the Prophet model provided in the embodiment of the present application trains the Prophet model based on the training time series data of the daily operation volume, wherein the training time series data of the daily operation volume includes a timestamp and a reference operation volume corresponding to the timestamp, and the Prophet model extracts the change trend, seasonal fluctuation and holiday effect from the training time series data of the daily operation volume and predicts it, and outputs the training prediction operation volume and the training residual term; based on the reference operation volume corresponding to the timestamp and the training prediction operation volume, the parameters of the Prophet model are adjusted to obtain the trained Prophet model. The trained Prophet model can be well used to predict the target operation volume corresponding to the target change day.
[0102] based on Figure 4 , Figure 5 A flowchart of a training method for a Gaussian Hidden Markov Model provided in one embodiment of the present application. Figure 5 As shown, the method of the embodiment of the present application may include:
[0103] S501. Obtain training time series data of the changed daily operation volume and training residual items. The training residual items are obtained when the Prophet model is trained by the training time series data of the daily operation volume.
[0104] In this step, the training time series data of the change day operation volume can be obtained based on the historical operation volume time series data. The training time series data of the change day operation volume includes the change day operation volume (such as the change day operation volume on Thursday or the change day operation volume on Saturday) and whether there is an emergency event. After obtaining the training time series data of the change day operation volume, in order to facilitate the Gaussian hidden Markov model to process the training time series data of the change day operation volume, the example of step S302 can be referred to, and the training time series data of the change day operation volume can be preprocessed to obtain the preprocessed training time series data of the change day operation volume, and the preprocessed training time series data of the change day operation volume can be used for the training of the Gaussian hidden Markov model. Among them, the preprocessing can include data cleaning, data format conversion processing and data normalization processing, and the data cleaning can include removing duplicate data, missing data processing and abnormal data processing.
[0105] Optionally, based on the historical operation volume time series data, the change day operation volume test time series data can be obtained to evaluate the performance of the Gaussian hidden Markov model finally trained.
[0106] For example, taking the change days being Thursday and Saturday as an example, a change day operation volume training data set can be obtained based on the historical operation volume time series data, and the change day operation volume training data set can be decomposed into Thursday change day operation volume training time series data, Thursday change day operation volume test time series data, Saturday change day operation volume training time series data and Saturday change day operation volume test time series data.
[0107] S502: Input the training time series data of the changed daily operation volume and the training residual term into the Gaussian Hidden Markov Model, and use the Baum-Welch algorithm to iteratively train the Gaussian Hidden Markov Model to obtain a trained Gaussian Hidden Markov Model.
[0108] Exemplarily, the operation amount prediction of the Gaussian hidden Markov model can be implemented by calling the hmmlearn library. In this step, first, a Gaussian hidden Markov model is established. When establishing the Gaussian hidden Markov model, it can specifically include the following three parts: (1) Evaluation part: Based on the given Gaussian hidden Markov model λ=(π,A,B) and the observation sequence O=(O1,O2,…Or) corresponding to the training residual term, the probability of the observation sequence P(O|λ) is generated, where A represents the state transfer matrix, B represents the observation matrix, and π represents the initial probability array of the state; the method for generating the observation sequence probability P(O|λ) is, for example, a forward-backward algorithm. The specific forward-backward algorithm can refer to the current related technology; (2) Decoding Part: Calculate the likelihood of the Gaussian hidden Markov model and construct the hidden Markov state, that is, find the optimal hidden variable sequence I = (i1, i2, ... ir) through the observation sequence O = (O1, O2, ... Or); the method for finding the optimal hidden variable sequence I = (i1, i2, ... ir) is, for example, the Viterbi algorithm. For the specific Viterbi algorithm, please refer to the current related technology; (3) Probability calculation part: Based on the probability P (O | λ) and the optimal hidden variable sequence I = (i1, i2, ... ir), adjust the parameters of the Gaussian hidden Markov model so that the Gaussian hidden Markov model λ can learn the observation sequence O, thereby obtaining the target model parameters. The method for adjusting the parameters of the Gaussian hidden Markov model is, for example, the Baum-Welch algorithm. For the specific Baum-Welch algorithm, please refer to the current related technology.
[0109] After the Gaussian hidden Markov model is built, the model parameters of the Gaussian hidden Markov model can be set. Optionally, the number of hidden states of the Gaussian hidden Markov model is 8 and the covariance type parameter of the observable state vector adopts a diagonal covariance matrix.
[0110] Exemplarily, the number of hidden states (n_components) of the Gaussian hidden Markov model can be set to 8, and the covariance type (covariance_type) parameter of the observable state vector can be set to the diagonal covariance matrix (diag). Then, the training time series data of the change day operation volume and the training residual items are input into the Gaussian hidden Markov model, and the Gaussian hidden Markov model is iteratively trained using the Baum-Welch algorithm to obtain a trained Gaussian hidden Markov model. Taking the change days as Thursday and Saturday as an example, the training time series data of Thursday's operation volume and the training residual items obtained from the Prophet model are input into the Gaussian hidden Markov model, and the prediction results of the Gaussian hidden Markov model with the change day as Thursday can be obtained; the training time series data of Saturday's operation volume and the training residual items obtained from the Prophet model are input into the Gaussian hidden Markov model, and the prediction results of the Gaussian hidden Markov model with the change day as Saturday can be obtained.
[0111] The training method of the Gaussian hidden Markov model provided in the embodiment of the present application is based on the training time series data of the change day operation volume and the training residual term, and the Baum-Welch algorithm is used to iteratively train the Gaussian hidden Markov model to obtain a trained Gaussian hidden Markov model; wherein the training residual term is obtained when training the Prophet model through the training time series data of the daily operation volume; the Gaussian hidden Markov model can learn and predict internal state transitions that cannot be directly observed, therefore, the trained Gaussian hidden Markov model can be better used to predict the target operation volume corresponding to the target change day.
[0112] In summary, the technical solution provided by this application has at least the following advantages:
[0113] (1) Fusion of the model structure in this application: The Prophet model is an open source time series prediction tool that can handle trends, seasonality, and holiday effects in time series data. It is very suitable for predicting time series data that is affected by multiple factors, such as changes in daily operation volume. The Gaussian Hidden Markov Model is a statistical model used to describe Markov processes with implicit unknown parameters. By combining the Prophet model and the Gaussian Hidden Markov Model, the implicit state of time series data and significant temporal patterns can be simultaneously mined, which is very effective for complex prediction tasks.
[0114] (2) Adaptability and accuracy of the model in this application: Compared with a single neural network model, such as LSTM or CNN, this application can better adapt to different data patterns and changes by combining the advantages of two different types of models, the Prophet model and the Gaussian Hidden Markov Model; the Prophet model can effectively handle external trends and cyclical changes, while the Gaussian Hidden Markov Model can fit and calculate residual terms, learn and predict internal state transitions that cannot be directly observed, and can mine the implicit state of time series data. The combination of the two can make the prediction results more accurate and reliable.
[0115] (3) The model's handling of outliers and the robustness of the model in this application: The Prophet model is designed to have a good ability to handle missing data and abnormal data, and this ability can be effectively combined with the state reasoning ability of the Gaussian hidden Markov model, thereby improving the overall model's adaptability to abnormal situations and the robustness of prediction.
[0116] (4) Complexity and operational efficiency of the implementation of this application: Although the combination of the Prophet model and the Gaussian hidden Markov model may lead to increased implementation complexity, this method can improve the accuracy of predictions; compared with deep learning models that require a large amount of computing resources, the combination of the Prophet model and the Gaussian hidden Markov model usually requires fewer computing resources, and the model training and prediction speeds are faster, which is suitable for business environments that require rapid response.
[0117] (5) Flexibility of application of this application: Since the parameters of the Prophet model and the Gaussian hidden Markov model are relatively independent, the adjustment and optimization of the model are more flexible; the parameters of the Prophet model and the Gaussian hidden Markov model can be independently adjusted according to specific business needs and data characteristics to achieve the best prediction effect.
[0118] The method for predicting personnel demand on the change day provided in the present application is based on a hybrid model of the Prophet model and the Gaussian hidden Markov model. In terms of the operation volume on the change day and the prediction of manpower demand on the change day, compared with the traditional neural network model (such as the LSTM deep learning model based on the dual attention mechanism for workload prediction), the present application can provide better timing analysis capabilities, higher exception handling capabilities and more accurate prediction accuracy, which is particularly suitable for prediction scenarios with dynamic changes and high data complexity in the field of operation and maintenance management.
[0119] The following is an embodiment of the device of the present application, which can be used to execute the embodiment of the method of the present application. For details not disclosed in the embodiment of the device of the present application, please refer to the embodiment of the method of the present application.
[0120] Figure 6This is a schematic diagram of the structure of a device for predicting personnel demand on a change date provided in an embodiment of the present application. Figure 6 As shown, the change day personnel demand prediction device 600 of the embodiment of the present application includes: a first acquisition module 601, a first prediction module 602, a second prediction module 603, a second acquisition module 604 and a determination module 605. Among them:
[0121] The first acquisition module 601 is used to obtain the time series data of the operation volume on weekdays before the target change date and the time series data of the operation volume on the change date in response to the personnel demand quantity prediction instruction for the target change date.
[0122] The first prediction module 602 is used to input the daily operation volume time series data into the Prophet model to predict the operation volume, and obtain the first predicted operation volume and target residual term output by the Prophet model.
[0123] The second prediction module 603 is used to input the time series data of the operation volume on the changed day and the target residual term into the Gaussian hidden Markov model to predict the operation volume, and obtain the second predicted operation volume output by the Gaussian hidden Markov model.
[0124] The second acquisition module 604 is used to obtain the target operation amount corresponding to the target change date based on the first predicted operation amount and the second predicted operation amount.
[0125] The determination module 605 is used to determine the target number of personnel required corresponding to the target change date according to the target operation volume.
[0126] In some embodiments, the second acquisition module 604 may be specifically used to: perform weighted summation on the first predicted operation amount and the second predicted operation amount to obtain the target operation amount corresponding to the target change date.
[0127] Optionally, the change-day personnel demand prediction device 600 may further include a training module (not shown in the figure), which is used to train and obtain a Prophet model in the following manner: obtaining daily operation volume training time series data, the daily operation volume training time series data including a timestamp and a reference operation volume corresponding to the timestamp; inputting the daily operation volume training time series data into the Prophet model for fitting, and obtaining the training predicted operation volume and training residual term corresponding to the timestamp output by the Prophet model; adjusting the parameters of the Prophet model based on the reference operation volume and the training predicted operation volume corresponding to the timestamp, and obtaining a trained Prophet model.
[0128] Optionally, the training module can also be used to train a Gaussian hidden Markov model in the following manner: obtain the training time series data and training residual items of the changed daily operation volume; input the training time series data and training residual items of the changed daily operation volume into the Gaussian hidden Markov model, and use the Baum-Welch algorithm to iteratively train the Gaussian hidden Markov model to obtain a trained Gaussian hidden Markov model.
[0129] In some embodiments, the number of hidden states of the Gaussian Hidden Markov Model is 8 and the covariance type parameters of the observable state vector adopt a diagonal covariance matrix.
[0130] In some embodiments, when the first prediction module 602 is used to input the daily operation volume time series data into the Prophet model for operation volume prediction, it can be specifically used to: preprocess the daily operation volume time series data to obtain preprocessed daily operation volume time series data, the preprocessing includes at least one of data cleaning, data format conversion processing and data normalization processing, and data cleaning includes at least one of removing duplicate data, missing data processing and abnormal data processing; input the preprocessed daily operation volume time series data into the Prophet model for operation volume prediction.
[0131] In some embodiments, when the second prediction module 603 is used to input the change day operation volume time series data and the target residual term into the Gaussian hidden Markov model for operation volume prediction, it can be specifically used for: preprocessing the change day operation volume time series data to obtain the preprocessed change day operation volume time series data; inputting the preprocessed change day operation volume time series data and the target residual term into the Gaussian hidden Markov model for operation volume prediction.
[0132] The personnel demand prediction device for the change date of the embodiment of the present application can be used to execute the technical solution of any of the above-mentioned method embodiments. Its implementation principle and technical effect are similar and will not be repeated here.
[0133] Figure 7 This is a schematic diagram of the structure of an electronic device provided by an embodiment of the present application. Figure 7 As shown, the electronic device 700 may include: at least one processor 701 and a memory 702 .
[0134] The memory 702 is used to store programs. Specifically, the programs may include program codes, and the program codes include computer-executable instructions.
[0135] The memory 702 may include a high-speed random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.
[0136] The processor 701 is used to execute the computer-executable instructions stored in the memory 702 to implement the change date personnel demand prediction method described in the aforementioned method embodiment. Among them, the processor 701 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application. Specifically, when implementing the change date personnel demand prediction method described in the aforementioned method embodiment, the electronic device may be, for example, an electronic device with processing functions such as a server.
[0137] Optionally, the electronic device 700 may further include a communication interface 703. In a specific implementation, if the communication interface 703, the memory 702 and the processor 701 are implemented independently, the communication interface 703, the memory 702 and the processor 701 may be interconnected through a bus and communicate with each other. The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc., but it does not mean that there is only one bus or one type of bus.
[0138] Optionally, in a specific implementation, if the communication interface 703, the memory 702 and the processor 701 are integrated on a chip, the communication interface 703, the memory 702 and the processor 701 can communicate through an internal interface.
[0139] The electronic device is used to execute the change day personnel demand prediction method in any of the aforementioned method embodiments, and its implementation principle and technical effect are similar and will not be repeated here.
[0140] The present application also provides a computer-readable storage medium, in which computer program instructions are stored. When a processor executes the computer program instructions, the above-mentioned method for predicting personnel demand on a change date is implemented.
[0141] The present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the solution of the above-mentioned method for predicting personnel demand on a change date.
[0142] The computer-readable storage medium mentioned above can 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. The readable storage medium can be any available medium that can be accessed by a general or special-purpose computer.
[0143] An exemplary readable storage medium is coupled to the processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in a dedicated integrated circuit. Of course, the processor and the readable storage medium can also exist as discrete components in the change day personnel demand forecasting device.
[0144] The division of units is only a logical function division, and there may be other divisions in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.
[0145] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0146] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
[0147] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory, a random access memory (RAM), a disk or an optical disk.
[0148] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, disk or optical disk and other media that can store program codes.
[0149] Finally, it should be noted that those skilled in the art will readily conceive of other embodiments of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any variations, uses or adaptations of the present application, which follow the general principles of the present application and include common knowledge or customary technical means in the art that are not disclosed in the present application, are not limited to the precise structures described above and shown in the drawings, and may be modified and changed in various ways without departing from the scope thereof. The scope of the present application is limited only by the appended claims.
Claims
1. A method for predicting personnel demand on a change date, characterized in that: include: In response to a personnel demand quantity forecasting instruction for a target change date, obtaining the time series data of the operation volume on weekdays before the target change date and the time series data of the operation volume on the change date; Inputting the daily operation volume time series data into the Prophet model to perform operation volume prediction, and obtaining a first predicted operation volume and a target residual term output by the Prophet model; Inputting the time series data of the operation amount on the changed day and the target residual term into a Gaussian hidden Markov model to predict the operation amount, and obtaining a second predicted operation amount output by the Gaussian hidden Markov model; Based on the first predicted operation amount and the second predicted operation amount, obtaining a target operation amount corresponding to the target change date; The target number of personnel required for the target change date is determined based on the target operation volume.
2. The method for predicting personnel demand on a change date according to claim 1, characterized in that: The obtaining, based on the first predicted operation amount and the second predicted operation amount, a target operation amount corresponding to the target change date includes: A weighted sum is performed on the first predicted operation amount and the second predicted operation amount to obtain a target operation amount corresponding to the target change date.
3. The method for predicting personnel demand on a change date according to claim 1, characterized in that: The Prophet model is trained in the following way: Acquire daily operation amount training time series data, wherein the daily operation amount training time series data includes a timestamp and a reference operation amount corresponding to the timestamp; Input the daily operation volume training time series data into the Prophet model for fitting, and obtain the training prediction operation volume and training residual term corresponding to the timestamp output by the Prophet model; Based on the reference operation amount and the training prediction operation amount corresponding to the timestamp, the parameters of the Prophet model are adjusted to obtain a trained Prophet model.
4. The method for predicting personnel demand on a change date according to claim 3, characterized in that: The Gaussian Hidden Markov Model is obtained by training in the following way: Obtaining the training time series data of the changed daily operation volume and the training residual item; The training time series data of the changed daily operation volume and the training residual term are input into a Gaussian hidden Markov model, and the Gaussian hidden Markov model is iteratively trained using a Baum-Welch algorithm to obtain a trained Gaussian hidden Markov model.
5. The method for predicting personnel demand on a change date according to claim 4, characterized in that: The number of hidden states of the Gaussian hidden Markov model is 8 and the covariance type parameter of the observable state vector adopts a diagonal covariance matrix.
6. The method for predicting personnel demand on a change date according to any one of claims 1 to 5, characterized in that: The step of inputting the daily operation volume time series data into the Prophet model to perform operation volume prediction includes: Preprocessing the daily operation volume time series data to obtain preprocessed daily operation volume time series data, wherein the preprocessing includes at least one of data cleaning, data format conversion processing and data normalization processing, and the data cleaning includes at least one of removing duplicate data, missing data processing and abnormal data processing; The pre-processed daily operation volume time series data is input into the Prophet model for operation volume prediction.
7. The method for predicting personnel demand on a change date according to claim 6, characterized in that: The step of inputting the time series data of the change day operation volume and the target residual term into a Gaussian hidden Markov model to perform operation volume prediction includes: Performing the preprocessing on the time series data of the change-day operation amount to obtain the preprocessed time series data of the change-day operation amount; The pre-processed time series data of the changed daily operation volume and the target residual term are input into a Gaussian hidden Markov model to perform operation volume prediction.
8. A device for predicting personnel demand on a change date, characterized in that: include: A first acquisition module is used to respond to the personnel demand quantity prediction instruction for the target change date, and acquire the normal day operation volume time series data before the target change date and the change day operation volume time series data; A first prediction module, used for inputting the daily operation volume time series data into a Prophet model to perform operation volume prediction, and obtaining a first predicted operation volume and a target residual term output by the Prophet model; A second prediction module is used to input the time series data of the operation volume on the change day and the target residual term into a Gaussian hidden Markov model to perform operation volume prediction, and obtain a second predicted operation volume output by the Gaussian hidden Markov model; A second acquisition module, configured to obtain a target operation amount corresponding to the target change date based on the first predicted operation amount and the second predicted operation amount; A determination module is used to determine the target number of personnel required corresponding to the target change date according to the target operation volume.
9. An electronic device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method for predicting personnel demand on a change date according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program instructions, and when the computer program instructions are executed, the method for predicting personnel demand on a change date according to any one of claims 1 to 7 is implemented.