Student score online monitoring method and system for processing high-dimensional features based on SVR model

Abstract

The invention discloses a student score online monitoring method and system for processing high-dimensional features based on an SVR model, and relates to the technical field of student score monitoring, and the method comprises the following steps: S1, data collection; s2, initial parameters are set; S3, a model is constructed and trained; step S4, feature selection and parameter optimization; and S5, carrying out online monitoring and feedback. According to the method, the improved breach algorithm is introduced to select and optimize the features and the model parameters, compared with a traditional method, the IDA algorithm continuously searches better solutions in a parameter space by simulating the breach process, and finally a parameter combination enabling the model prediction error to be minimum is found, so that the prediction accuracy and efficiency of the SVR model are improved, and the prediction efficiency of the SVR model is improved. The calculation complexity of the model is reduced, the parameters of the SVR model and the feature selection process are optimized, the adaptability of the model is improved, real-time monitoring and feedback of student scores are achieved, the real learning condition of students is accurately reflected, and the requirement for real-time monitoring is met.

Description

Technical Field

[0001] The present invention relates to the technical field of student performance monitoring, and in particular to a student performance online monitoring method and system based on SVR model processing of high-dimensional features. Background Art

[0002] The online student performance monitoring method and system mainly covers a series of processes and tools for recording, managing, analyzing and querying student performance using Internet technology. The online student performance monitoring method and system provide teachers with convenient and efficient performance management tools, which help to improve teaching quality and student learning outcomes. The monitoring of student performance mostly relies on traditional statistical models or simple data-driven models. These methods often have problems such as high computational complexity and insufficient prediction accuracy when processing high-dimensional feature data, thus affecting the accuracy and real-time performance of performance prediction. Existing student performance prediction methods mainly rely on models such as multivariate linear regression, logistic regression, decision trees or simple neural networks. These methods lack an effective mechanism for feature selection, resulting in redundant features that have a negative impact on model performance.

[0003] The defects of the performance monitoring method and system in the prior art are:

[0004] 1. Patent document CN114861832A discloses a student performance analysis method and system based on an elite genetic clustering algorithm. The poor adaptability of the model in processing nonlinear relationships and high-dimensional data such as student performance prediction will cause the prediction results to deviate from reality and fail to accurately reflect the students' actual learning situation.

[0005] 2. Patent document CN114492156A discloses a student performance prediction method, system, computer and medium based on big data. This performance prediction method requires a large amount of calculation to find the optimal solution when processing high-dimensional data, thereby consuming a large amount of computing resources and time. The high computational complexity reduces the prediction efficiency and cannot meet the real-time monitoring requirements.

[0006] 3. Patent document CN117763361B discloses a student performance prediction method and system based on artificial intelligence. The data-driven model in this student performance prediction system is insufficient in feature selection and parameter optimization. Insufficient parameter optimization may cause the model to overfit or underfit, further reducing the prediction performance and making it impossible to achieve real-time monitoring and feedback of student performance. Summary of the invention

[0007] The purpose of the present invention is to provide a method and system for online monitoring of student performance based on SVR model processing high-dimensional features, so as to solve the problems raised in the above background technology.

[0008] To achieve the above-mentioned purpose, the present invention provides the following technical solution: a method for online monitoring of student performance based on SVR model processing high-dimensional features, comprising the following steps: step S1, data collection; step S2, initial parameter setting; step S3, model construction and training; step S4, feature selection and parameter optimization; step S5, online monitoring and feedback;

[0009] The feature selection and parameter optimization steps are as follows: the IDA algorithm simultaneously screens high-dimensional features and SVR parameters. The IDA algorithm continuously optimizes feature subsets and parameters by simulating the duel process between individuals, and searches for the optimal feature and parameter combination that minimizes the model prediction error.

[0010] Preferably, the data collection steps are as follows: through multiple channels such as the school's information management system, online learning platform, classroom interaction system, etc., the students' gender, age, grade, basic information of the major and attendance rate, number of homework submissions, number of classroom interactions, online learning time, library visit records and learning behavior characteristics are collected in real time, and the students' basic information and learning behavior data are pre-processed.

[0011] Preferably, the initial parameter setting step is as follows: dividing the preprocessed data into training data and test data, initializing feature subsets and SVR parameters.

[0012] Preferably, the model construction and training steps are as follows: when constructing the SVR model, a suitable radial basis function RBF is selected and corresponding penalty parameter C, kernel width γ, and allowable error parameter are set, and the SVR model is trained.

[0013] Preferably, the online monitoring and feedback steps are as follows: use the deployed monitoring system to predict the grades of new data, find the optimal solution for feature subsets and SVR model parameters, and output the online monitoring results of students' grades. The prediction results are fed back to teachers and students in real time to help them understand students' learning status and potential problems, and provide students with personalized learning suggestions based on the prediction results.

[0014] Preferably, an algorithm formula is used to normalize the initial sample data, and the algorithm formula is as follows:

[0015]

[0016] where a ij is the sample data, a imin with a imax are the minimum and maximum values ​​of the learning behavior characteristics of the i-th student.

[0017] Preferably, the initial parameters of the support vector regression model and the improved duel algorithm are set, including the mutation probability P m =0.1, learning probability Pl =0.8, improved probability P c = 0.1, step size coefficient λ = 5, the search range of penalty parameter C is C∈[10 -3 ,10 3 ], the search range of kernel function bandwidth γ is γ∈[10 -6 ,10 1 ], the search range of the allowable error ε∈[10 -3 ,10 -1 ], current iteration number t = 1, maximum iteration number t max .

[0018] Preferably, in step S3, the model construction and training specifically includes the following steps:

[0019] Step S31, initialize the population and encode the initial solution. The encoding includes two parts: feature subset N and support vector machine parameters (C, γ, ε). The i-th solution can be expressed as:

[0020]

[0021] Step S32, using mutation operation on each solution to generate a new solution;

[0022] Step S33, performing decoding operation and calculating fitness;

[0023] Step S34: randomly generate a competition table, and make the current solution compete one-on-one according to the competition table to determine the winner and the loser;

[0024] Step S35, the loser replaces a part of its binary array with the value of the winner's binary array to improve its fitness, and the winner changes the value of its binary array to the new value to improve its fitness;

[0025] Step S36, update the optimal solution, set t = t + 1, and determine t <t max Is it true? If so, execute step S37; otherwise, end the search and output the optimal solution;

[0026] Step S37, increase the number of iterations by one, and return to step S32.

[0027] Preferably, the IDA algorithm performs feature selection and model parameter optimization, and uses the Logistic equation to construct a chaotic sequence:

[0028] x(t+1)=μx(t)(1-x(i))i=0,1,2,…

[0029] Among them, μ is the control parameter. When 0 < x(0) < 1 and μ = 4, the Logistic equation is in a completely chaotic state. At this time, x(i) is chaotic within the interval (0, 1). Given the initial value x(0) ∈ (0, 1), the time series x(1), x(2),... can be generated;

[0030] Define the adaptive lucky coefficient:

[0031]

[0032] Among them, c is the lucky coefficient of the current iteration number t, and λ is the step size.

[0033] An online monitoring system for students' grades that processes high-dimensional features based on the SVR model, applicable to an online monitoring method for students' grades that processes high-dimensional features based on the SVR model, including a data transmission module, a parameter setting module, a model training module, an intelligent decision-making module, and an abnormal alarm module.

[0034] Preferably, the data transmission module is used to execute step S1, collect the basic information and learning behavior data of students, and preprocess the collected data;

[0035] The parameter setting module is used to execute step S2, set the parameters required for the monitoring system, including mutation probability, learning probability, improvement probability, penalty parameter, kernel function bandwidth, and allowable error;

[0036] The model training module is used to execute step S3, import the learning sample data (X train , Y train ), use the learning sample data to train the support vector machine regression model, initialize the model Y train =

[0037] SVR(X train ) and adopt the improved IDA algorithm to simultaneously select the learning behavior features and optimize the model parameters, and find the optimal feature subset and the optimal solution of the model parameters;

[0038] The intelligent decision-making module imports the online monitoring sample data X test , uses the trained model to test the grades of the online monitoring samples, Y test = SVR(X test ), and outputs the grade monitoring result Y test ;

[0039] The abnormal alarm module quickly judges the student grade monitoring results, notifies the relevant teachers of the students with abnormal grades in the first time, and records, views, and tracks the abnormal students.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] 1. The present invention introduces an improved duel algorithm to select and optimize features and model parameters to improve the prediction accuracy of the model. Compared with traditional methods, the IDA algorithm continuously searches for better solutions in the parameter space by simulating the duel process, and finally finds the parameter combination that minimizes the model prediction error, thereby improving the prediction accuracy and efficiency of the SVR model and reducing the computational complexity of the model. The parameters and feature selection process of the SVR model are optimized to improve the adaptability of the model and realize real-time monitoring and feedback of student performance, preventing the prediction results from deviating from reality due to poor model adaptability, accurately reflecting the students' actual learning situation and meeting the needs of real-time monitoring. Teachers can understand the students' learning situation in a timely manner and take intervention measures to avoid missing the best time for teaching adjustment.

[0042] 2. The present invention introduces strategies such as chaotic sequence initialization population, roulette, and adaptive luck coefficient. The IDA algorithm can quickly converge to the optimal solution while ensuring the quality of the solution. These optimization strategies enable the IDA algorithm to maintain a high efficiency when processing high-dimensional feature data, further reducing the computational complexity of the model. Chaos improves the diversity of the population, allowing the algorithm to perform a more detailed search near the optimal solution, increasing the diversity of the loser's improvement to avoid the influence of a single winner on the loser's improvement, and generating neighborhood solutions of the local optimal solution during the iteration process to help the current solution escape the local optimal solution, thereby improving the global search capability of the algorithm. Multiple non-overlapping strategies jointly guarantee the prediction accuracy and convergence speed of the algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a flow chart of the online monitoring method of student performance of the present invention;

[0044] Figure 2 The improved duel algorithm flow chart of the present invention;

[0045] Figure 3 It is a schematic diagram of the structure of the online monitoring system for student performance of the present invention. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] In the description of the present invention, it should be noted that the terms "upper", "lower", "inner", "outer", "front end", "rear end", "two ends", "one end", "the other end" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance.

[0048] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "provided with", "connected", etc. should be understood in a broad sense. For example, "connected" can be a fixed connection or a movable connection, or a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0049] Example 1, see Figure Figure 1 , an embodiment provided by the present invention: a method for online monitoring of student performance based on SVR model processing high-dimensional features, comprising the following steps:

[0050] Step S1, data collection;

[0051] Step S2: Initial parameter setting:

[0052] Step S3: model construction and training;

[0053] Step S4: feature selection and parameter optimization;

[0054] Step S5: Online monitoring and feedback.

[0055] Example 2, see Figure Figure 1 ,This embodiment is based on the above embodiment, and the data collection steps are as follows: through the school’s information management system, online learning platform, classroom interaction system and other channels, the students’ gender, age, grade, and basic information of their majors and attendance rate, the number of homework submissions, the number of classroom interactions, online learning time, library visit records and learning behavior characteristics are collected in real time, and the students’ basic information and learning behavior data are preprocessed;

[0056] The initial parameter setting steps are as follows: divide the preprocessed data into training data and test data, initialize the feature subset and SVR parameters;

[0057] The steps of model construction and training are as follows: when constructing the SVR model, select the appropriate radial basis function RBF and set the corresponding penalty parameter C, kernel width γ, and allowable error parameter, and train the SVR model;

[0058] The steps of online monitoring and feedback are as follows: Use the deployed monitoring system to predict the grades of new data, find the optimal solution for feature subsets and SVR model parameters, and output the online monitoring results of students' grades. The prediction results are fed back to teachers and students in real time to help them understand students' learning status and potential problems, and provide personalized learning suggestions for students based on the prediction results;

[0059] Example 3, see Figure Figure 1 and Figure 2 ,This embodiment is based on the above embodiment, and the data collection steps are as follows: through the school’s information management system, online learning platform, classroom interaction system and other channels, the students’ gender, age, grade, and basic information of their majors and attendance rate, the number of homework submissions, the number of classroom interactions, online learning time, library visit records and learning behavior characteristics are collected in real time, and the students’ basic information and learning behavior data are preprocessed;

[0060] The initial parameter setting steps are as follows: divide the preprocessed data into training data and test data, initialize the feature subset and SVR parameters;

[0061] The algorithm formula is used to normalize the initial sample data. The algorithm formula is as follows:

[0062]

[0063] where a ij is the sample data, a imin with a imax are the minimum and maximum values ​​of the learning behavior characteristics of the i-th student.

[0064] Example 4, see Figure Figure 1 and Figure 2 ,This embodiment is based on the above embodiment, and the data collection steps are as follows: through the school’s information management system, online learning platform, classroom interaction system and other channels, the students’ gender, age, grade, and basic information of their majors and attendance rate, the number of homework submissions, the number of classroom interactions, online learning time, library visit records and learning behavior characteristics are collected in real time, and the students’ basic information and learning behavior data are preprocessed;

[0065] The initial parameter setting steps are as follows: divide the preprocessed data into training data and test data, initialize the feature subset and SVR parameters;

[0066] The steps of model construction and training are as follows: when constructing the SVR model, select the appropriate radial basis function RBF and set the corresponding penalty parameter C, kernel width γ, and allowable error parameter to train the SVR model;

[0067] The algorithm formula is used to normalize the initial sample data. The algorithm formula is as follows:

[0068]

[0069] where a ij is the sample data, a imin with a imax is the minimum and maximum value of the learning behavior characteristics of the i-th student;

[0070] Set the initial parameters of the support vector regression model and the improved duel algorithm, including the mutation probability P m =0.1, learning probability P l =0.8, improved probability P c = 0.1, step size coefficient λ = 5, the search range of penalty parameter C is C∈[10 -3 ,10 3 ], the search range of kernel function bandwidth γ is γ∈[10 -6 ,10 1 ], the search range of the allowable error ε∈[10 -3 ,10 -1 ], current iteration number t = 1, maximum iteration number t max ;

[0071] In step S3, model building and training, the following steps are specifically included:

[0072] Step S31, initialize the population and encode the initial solution. The encoding includes two parts: feature subset N and support vector machine parameters (C, γ, ε). The i-th solution can be expressed as:

[0073]

[0074] Step S32, using mutation operation on each solution to generate a new solution;

[0075] Step S33, performing decoding operation and calculating fitness;

[0076] Step S34: randomly generate a competition table, and make the current solution compete one-on-one according to the competition table to determine the winner and the loser;

[0077] Step S35, the loser replaces a part of its binary array with the value of the winner's binary array to improve its fitness, and the winner changes the value of its binary array to the new value to improve its fitness;

[0078] Step S36, update the optimal solution, set t = t + 1, and determine t <t max Is it true? If not, end the search and output the optimal solution.

[0079] Example 5, see Figure Figure 1 and Figure 2 ,This embodiment is based on the above embodiment, and the data collection steps are as follows: through the school’s information management system, online learning platform, classroom interaction system and other channels, the students’ gender, age, grade, and basic information of their majors and attendance rate, the number of homework submissions, the number of classroom interactions, online learning time, library visit records and learning behavior characteristics are collected in real time, and the students’ basic information and learning behavior data are preprocessed;

[0080] The initial parameter setting steps are as follows: divide the preprocessed data into training data and test data, initialize the feature subset and SVR parameters;

[0081] The steps of model construction and training are as follows: when constructing the SVR model, select the appropriate radial basis function RBF and set the corresponding penalty parameter C, kernel width γ, and allowable error parameter to train the SVR model;

[0082] The algorithm formula is used to normalize the initial sample data. The algorithm formula is as follows:

[0083]

[0084] where a ij is the sample data, a imin with a imax are the minimum and maximum values ​​of the learning behavior characteristics of the i-th student.

[0085] Set the initial parameters of the support vector regression model and the improved duel algorithm, including the mutation probability P m =0.1, learning probability P l =0.8, improved probability P c = 0.1, step size coefficient λ = 5, the search range of penalty parameter C is C∈[10 -3 ,10 3 ], the search range of kernel function bandwidth γ is γ∈[10 -6 ,10 1 ], the search range of the allowable error ε∈[10 -3 ,10 -1 ], current iteration number t = 1, maximum iteration number t max .

[0086] In step S3, model building and training, the following steps are specifically included:

[0087] Step S31, initialize the population and encode the initial solution. The encoding includes two parts: feature subset N and support vector machine parameters (C, γ, ε). The i-th solution can be expressed as:

[0088]

[0089] Step S32, using mutation operation on each solution to generate a new solution;

[0090] Step S33, performing decoding operation and calculating fitness;

[0091] Step S34: randomly generate a competition table, and make the current solution compete one-on-one according to the competition table to determine the winner and the loser;

[0092] Step S35, the loser replaces a part of its binary array with the value of the winner's binary array to improve its fitness, and the winner changes the value of its binary array to the new value to improve its fitness;

[0093] Step S36, update the optimal solution, set t = t + 1, and determine t <t max Is it established? If yes, execute step S37;

[0094] Step S37, increase the number of iterations by one, and return to step S32.

[0095] Example 6, see Figure Figure 1 and Figure 2 ,This embodiment is based on the above embodiment, and the data collection steps are as follows: through the school’s information management system, online learning platform, classroom interaction system and other channels, the students’ gender, age, grade, and basic information of their majors and attendance rate, the number of homework submissions, the number of classroom interactions, online learning time, library visit records and learning behavior characteristics are collected in real time, and the students’ basic information and learning behavior data are preprocessed;

[0096] The initial parameter setting steps are as follows: divide the preprocessed data into training data and test data, initialize the feature subset and SVR parameters;

[0097] The steps of model construction and training are as follows: when constructing the SVR model, select the appropriate radial basis function RBF and set the corresponding penalty parameter C, kernel width γ, and allowable error parameter to train the SVR model;

[0098] The steps of feature selection and parameter optimization are as follows: The IDA algorithm simultaneously screens high-dimensional features and SVR parameters. The IDA algorithm continuously optimizes feature subsets and parameters by simulating the duel process between individuals, and finds the optimal feature and parameter combination that minimizes the model prediction error;

[0099] The steps of online monitoring and feedback are as follows: Use the deployed monitoring system to predict the grades of new data, find the optimal solution for feature subsets and SVR model parameters, and output the online monitoring results of students' grades. The prediction results are fed back to teachers and students in real time to help them understand students' learning status and potential problems, and provide personalized learning suggestions for students based on the prediction results;

[0100] The algorithm formula is used to normalize the initial sample data. The algorithm formula is as follows:

[0101]

[0102] where a ij is the sample data, a imin with a imax are the minimum and maximum values ​​of the learning behavior characteristics of the i-th student.

[0103] Set the initial parameters of the support vector regression model and the improved duel algorithm, including the mutation probability P m =0.1, learning probability P l =0.8, improved probability P c = 0.1, step size coefficient λ = 5, the search range of penalty parameter C is C∈[10 -3 ,10 3 ], the search range of kernel function bandwidth γ is γ∈[10 -6 ,10 1 ], the search range of the allowable error ε∈[10 -3 ,10 -1 ], current iteration number t = 1, maximum iteration number t max ;

[0104] In step S3, model building and training, the following steps are specifically included:

[0105] Step S31, initialize the population and encode the initial solution. The encoding includes two parts: feature subset N and support vector machine parameters (C, γ, ε). The i-th solution can be expressed as:

[0106]

[0107] Step S32, using mutation operation on each solution to generate a new solution;

[0108] Step S33, performing decoding operation and calculating fitness;

[0109] Step S34, randomly generate a competition table, and make the current solution compete one-on-one according to the competition table to determine the winner and the loser;

[0110] Step S35: The loser improves its fitness by replacing a part of its binary array with the binary array value of the winner, and the winner changes the value of its binary array to a new value to improve its fitness;

[0111] Step S36: Update the optimal solution, let t = t + 1, and determine whether t < t max holds. If it holds, execute Step S37; otherwise, end the search and output the optimal solution;

[0112] Step S37: Increment the iteration count by one and return to Step S32.

[0113] The IDA algorithm performs feature selection and optimization of model parameters, and constructs a chaotic sequence using the Logistic equation:

[0114] x(t + 1) = μx(t)(1 - x(i)), i = 0, 1, 2, …

[0115] where μ is the control parameter. When 0 < x(0) < 1 and μ = 4, the Logistic equation is in a completely chaotic state. At this time, x(i) is chaotic in the interval (0, 1). Given the initial value x(0) ∈ (0, 1), the time series x(1), x(2), … can be generated;

[0116] Define the adaptive lucky coefficient:

[0117]

[0118] where c is the lucky coefficient for the current iteration count t, and λ is the step size.

[0119] Example 7, refer to Figure Figure 1 、 Figure 2 and Figure 3 This example is based on the above example. An example provided by the present invention: An online monitoring system for students' grades based on the SVR model to process high-dimensional features, including a data transmission module, a parameter setting module, a model training module, an intelligent decision-making module, and an abnormal alarm module;

[0120] The data transmission module is used to execute Step S1, collect the basic information and learning behavior data of students, and preprocess the collected data;

[0121] The parameter setting module is used to execute Step S2, set the parameters required for the monitoring system, including the mutation probability, learning probability, improvement probability, penalty parameter, kernel function bandwidth, and allowable error;

[0122] The model training module is used to execute Step S3, import the learning sample data (X train , Y train), use the learning sample data to train the support vector machine regression model and initialize the model Y train =SVR(X train ) and adopts the improved IDA algorithm to simultaneously select learning behavior features and optimize model parameters to find the optimal feature subset and the optimal solution for model parameters;

[0123] Intelligent decision-making module imports online monitoring sample data X test , use the trained model to test the performance of online monitoring samples, Y test =SVR(X test ), output the performance monitoring result Y test ;

[0124] The abnormal alarm module makes a quick judgment on the student performance monitoring results, notifies the relevant teachers of students with abnormal performance as soon as possible, and records, views and tracks the abnormal students.

[0125] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A method for online monitoring of student performance based on SVR model processing high-dimensional features, characterized in that: It includes the following steps: Step S1, data collection; Step S2, initial parameter setting; Step S3, model construction and training; Step S4, feature selection and parameter optimization; Step S5, online monitoring and feedback; The steps of the feature selection and parameter optimization are as follows: The IDA algorithm simultaneously screens high-dimensional features and SVR parameters. The IDA algorithm continuously optimizes the feature subset and parameters by simulating the duel process between individuals, and searches for the optimal features and parameter combinations that minimize the model prediction error.

2. The method for online monitoring of student performance based on SVR model processing of high-dimensional features according to claim 1 is characterized in that: The steps of the data collection are as follows: Through multiple channels such as the school's information management system, online learning platform, and classroom interaction system, real-time collect the basic information of students' gender, age, grade, major and learning behavior characteristics such as attendance rate, number of homework submissions, number of classroom interactions, online learning time, and library visit records, and preprocess the basic information and learning behavior data of students.

3. The method for online monitoring of student performance based on SVR model processing of high-dimensional features according to claim 1 is characterized in that: The steps of the initial parameter setting are as follows: Divide the preprocessed data into training data and test data, and initialize the feature subset and SVR parameters.

4. The method for online monitoring of student performance based on SVR model processing of high-dimensional features according to claim 1 is characterized in that: The steps of the model construction and training are as follows: When constructing the SVR model, select an appropriate radial basis function RBF and set the corresponding penalty parameter C, kernel width γ, and allowable error parameter, and train the SVR model.

5. The method for online monitoring of student performance based on SVR model processing of high-dimensional features according to claim 1 is characterized in that: The steps of the online monitoring and feedback are as follows: Use the deployed monitoring system to predict the grades of new data, find the optimal solutions for the feature subset and SVR model parameters, and output the online monitoring results of students' grades. The prediction results are real-time fed back to teachers and students to help them understand the students' learning status and potential problems, and provide personalized learning suggestions for students according to the prediction results.

6. The method for online monitoring of student performance based on SVR model processing of high-dimensional features according to claim 2 is characterized in that: The initial sample data is normalized using an algorithm formula, and the algorithm formula is as follows: where a ij is the sample data, a imin with a imax are the minimum and maximum values ​​of the learning behavior characteristics of the i-th student.

7. The method for online monitoring of student performance based on SVR model processing of high-dimensional features according to claim 4 is characterized in that: Set the initial parameters of the support vector regression model and the improved duel algorithm, including the mutation probability P m =0.1, learning probability P l =0.8, improved probability P c = 0.1, step size coefficient λ = 5, the search range of penalty parameter C is C∈[10 -3 ,10 3 ], the search range of kernel function bandwidth γ is γ∈[10 -6 ,10 1 ], the search range of the allowable error ε∈[10 -3 ,10 -1 ], current iteration number t = 1, maximum iteration number t max .

8. The method for online monitoring of student performance based on SVR model processing of high-dimensional features according to claim 7 is characterized in that: In step S3, the model construction and training specifically include the following steps: Step S31, initialize the population, encode the initial solution, and the encoding includes two parts: the feature subset N and the support vector machine parameters (C, γ, ε). The i-th solution can be expressed as: Step S32, perform a mutation operation on each solution to generate a new solution; Step S33, perform a decoding operation and calculate the fitness; Step S34, randomly generate a competition table, and let the current solution compete one-on-one according to the competition table to determine the winner and the loser; Step S35, the loser improves the fitness by replacing a part of its binary array value with the binary array value of the winner, and the winner changes the value of its binary array to a new value to improve the fitness; Step S36, update the optimal solution, set t = t + 1, and determine t <t max Is it true? If so, execute step S37; otherwise, end the search and output the optimal solution; Step S37, increment the iteration count by one, and return to step S32.

9. The method for online monitoring of student performance based on SVR model processing of high-dimensional features according to claim 1 is characterized in that: The IDA algorithm performs feature selection and optimization of model parameters, and constructs a chaotic sequence using the Logistic equation: x(t + 1) = μx(t)(1 - x(i)) i = 0, 1, 2,... Where μ is the control parameter. When 0 < x(0) < 1 and μ = 4, the Logistic equation is in a completely chaotic state. At this time, x(i) is chaotic in the interval (0, 1). Given the initial value x(0) ∈ (0, 1), a time series x(1), x(2),... can be generated; Define the adaptive lucky coefficient: Among them, c is the lucky coefficient of the current iteration number t, and λ is the step size.

10. A student performance online monitoring system based on SVR model processing high-dimensional features, applicable to the student performance online monitoring method based on SVR model processing high-dimensional features according to any one of claims 1 to 9, characterized in that: It includes data transmission module, parameter setting module, model training module, intelligent decision-making module and abnormal alarm module.

11. The student performance online monitoring system based on SVR model processing high-dimensional features according to claim 10 is characterized by: The data transmission module is used to execute step S1, collect students' basic information and learning behavior data, and pre-process the collected data; The parameter setting module is used to execute step S2 to set the parameters required for the monitoring system, including mutation probability, learning probability, improvement probability, penalty parameter, kernel function bandwidth and allowable error; The model training module is used to execute step S3, importing learning sample data (X train ,Y train ), use the learning sample data to train the support vector machine regression model and initialize the model Y train =SVR(X train ) and adopts the improved IDA algorithm to simultaneously select learning behavior features and optimize model parameters to find the optimal feature subset and the optimal solution for model parameters; The intelligent decision-making module imports online monitoring sample data X test , use the trained model to test the performance of online monitoring samples, Y test =SVR(X test ), output the performance monitoring result Y test ; The abnormal alarm module makes a quick judgment on the student performance monitoring results, notifies the relevant teachers of students with abnormal performance as soon as possible, and records, views and tracks the abnormal students.

Images