A method and system for predicting student academic performance based on artificial intelligence

Abstract

The present invention discloses a method and system for predicting student academic performance based on artificial intelligence, and the method includes data collection, data preprocessing, learning behavior analysis, performance analysis prediction and student academic performance prediction. The present invention relates to the field of performance prediction technology, and specifically refers to a method and system for predicting student academic performance based on artificial intelligence. The present scheme adopts a method for predicting student academic performance by combining learning behavior with learning performance, through learning behavior characteristics at multiple levels and multiple dimensions; adopts an improved convolutional neural network combined with nearest neighbor clustering for learning behavior analysis, and by clustering learning behavior, improves the in-depth mining and feature extraction of common features and individual features; adopts a bidirectional long short-term memory neural network combined with stacked feature encoding and graph feature extraction for performance analysis prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of performance prediction, and in particular to a method and system for predicting student academic performance based on artificial intelligence. Background Art

[0002] The AI-based student academic performance prediction method uses machine learning and data analysis technology to mine multi-dimensional data such as students' historical grades, learning behaviors, and course participation, thereby predicting their future academic performance. This method can help educators identify students with learning difficulties, develop targeted tutoring measures, improve teaching effectiveness, and provide data support for schools' teaching decisions, thereby promoting the improvement of education quality.

[0003] However, in the existing intelligent prediction methods of student grades, there is a technical problem that the existing intelligent systems often focus on the simple prediction of grade values, but this prediction method often ignores the students' own active learning behavior and various other complex factors in the academic process; in the existing learning behavior analysis methods, there is a technical problem that learning behavior has both discrete individual characteristics and common learning behavior characteristics, so it is necessary to consider complex global characteristics and to mine similarity characteristics for the same type of learning behavior, and the existing technology ignores the common characteristics in the process of learning behavior analysis; in the existing grade analysis and prediction, there is a technical problem that the existing technology lacks a method to further predict student grades by combining the results of learning behavior analysis, and the specific use of the results of learning behavior analysis is not limited to simple feature input, but should also be used as a related factor for grade change and trend prediction. Summary of the invention

[0004] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a method and system for predicting student academic performance based on artificial intelligence. In view of the technical problem that in the existing intelligent prediction methods for student performance, the existing intelligent systems often focus on the simple numerical prediction of performance, but this prediction method often ignores the students' own active learning behavior and various other complex factors in the academic process, this solution creatively adopts a method for predicting student academic performance by combining learning behavior with learning performance, and deeply explores the potential reasons for fluctuations and changes in students' academic performance through learning behavior characteristics at multiple levels and dimensions, thereby improving the prediction depth of student performance prediction; in view of the fact that in the existing learning behavior analysis methods, learning behaviors have both discrete individual characteristics and common learning behavior characteristics, and therefore it is necessary to consider complex global characteristics and also to mine similarity characteristics for the same type of learning behaviors, while the prior art ignores the technical problem of common characteristics in the process of learning behavior analysis, this solution creatively adopts a method combining nearest neighbor aggregation to predict student academic performance. The improved convolutional neural network of the class is used to analyze learning behavior. By clustering learning behaviors, the in-depth mining and feature extraction of common features and individual features are improved, and the extraction depth of learning behavior features is further optimized through mixed pooling, which provides help for subsequent academic performance prediction; in the existing performance analysis and prediction, there is a lack of methods in the existing technology to further predict students' grades by combining the results of learning behavior analysis, and the specific use of the results of learning behavior analysis is not limited to simple feature input, but should also be used as a related factor for performance change and trend prediction. This solution creatively uses a bidirectional long short-term memory neural network that combines stacked feature encoding and graph feature extraction to perform performance analysis and prediction, optimizes feature extraction through multiple feature extraction methods, and corrects and adjusts the prediction results by taking learning behavior as a weight value. Finally, a bidirectional long short-term memory structure is used to improve the processing capability of time series data, and the prediction of students' academic performance combined with learning behavior analysis is realized, which improves the overall usability and reliability of the method.

[0005] The technical solution adopted by the present invention is as follows: The present invention provides a method for predicting student academic performance based on artificial intelligence, the method comprising the following steps:

[0006] Step S1: data collection;

[0007] Step S2: data preprocessing;

[0008] Step S3: learning behavior analysis;

[0009] Step S4: performance analysis and prediction;

[0010] Step S5: Prediction of students’ academic performance.

[0011] Furthermore, in step S1, the data collection is used to collect the original data set for predicting students' academic performance, specifically, to obtain the original data set for predicting students' academic performance through student academic survey and teaching system data collection;

[0012] The original data set for predicting student academic performance specifically includes historical performance data, study habit data, class participation data, social activity data, and mental health data.

[0013] Furthermore, in step S2, the data preprocessing is used to optimize the original data, specifically, to perform basic preprocessing and data enhancement operations on the original data set of student academic performance prediction to obtain enhanced student academic performance prediction data;

[0014] The enhanced student academic performance prediction data specifically includes an enhanced learning behavior analysis data set and an enhanced performance analysis prediction data set.

[0015] Further, in step S3, the learning behavior analysis is used to perform learning behavior analysis of the student's academic process and provide a reference for subsequent performance prediction, specifically, based on the enhanced student academic performance prediction data, an improved convolutional neural network combined with nearest neighbor clustering is used to perform learning behavior analysis to obtain a learning behavior analysis reference data set;

[0016] The improved convolutional neural network combined with nearest neighbor clustering specifically includes a nearest neighbor clustering subnet and an improved convolutional subnet;

[0017] The step of using the improved convolutional neural network combined with nearest neighbor clustering to perform learning behavior analysis to obtain a learning behavior analysis reference data set includes:

[0018] Step S31: construct a nearest neighbor clustering subnetwork to classify students according to their learning behaviors and aggregate similar learning behavior patterns. Specifically, the nearest neighbor clustering is performed through K-means calculation to obtain clustered student behavior data. The calculation formula is:

[0019] ;

[0020] In the formula, J is the K-means clustering objective function, which is used to calculate the clustered student behavior data, K is the total number of cluster centers, k is the cluster center index, N is the total number of cluster samples, n is the cluster sample index, is the nth input data sample corresponding to the kth cluster center, is the mean of the kth cluster center;

[0021] Step S32: constructing an improved convolutional subnet, specifically constructing a multi-layer convolutional layer, and using a nonlinear activation function and an improved comprehensive pooling operation to perform feature processing, and obtaining convolution feature data based on the clustered student behavior data;

[0022] Step S33: training the learning behavior analysis model, specifically, training the learning behavior analysis model through the nearest neighbor clustering subnet and the improved convolution subnet, and using the cross entropy loss function as the model loss function to obtain the learning behavior analysis model Model BA ;

[0023] Step S34: learning behavior analysis, specifically using the learning behavior analysis model Model BA , based on the enhanced student academic performance prediction data, learning behavior analysis is performed to obtain a learning behavior analysis reference data set.

[0024] Further, in step S4, the performance analysis prediction is used to perform performance analysis prediction in combination with the results of the student behavior analysis, specifically, based on the enhanced student academic performance prediction data and the learning behavior analysis reference data set, a bidirectional long short-term memory neural network combining stacked feature coding and graph feature extraction is used to perform performance analysis prediction to obtain a performance analysis prediction data set;

[0025] The bidirectional long short-term memory neural network combining stacked feature encoding and graph feature extraction specifically includes a stacked feature encoder, a graph feature extraction subnet, a bidirectional long short-term memory subnet and a prediction output layer;

[0026] The step of using a bidirectional long short-term memory neural network combining stacked feature encoding and graph feature extraction to perform performance analysis and prediction to obtain a performance analysis and prediction data set includes:

[0027] Step S41: constructing a stacked feature encoder, specifically constructing a multi-layer autoencoder for feature learning, and extracting the enhanced student academic performance prediction data and the learning behavior analysis reference data set as feature data and performing data integration to obtain stacked autoencoder data;

[0028] Step S42: construct a graph feature extraction subnet, specifically, based on the stacked autoencoder data, construct a graph convolutional neural network as a graph feature extraction subnet, and integrate the learning behavior analysis reference data set to generate a learning behavior offset weight, and generate a graph convolution feature. The calculation formula is:

[0029] ;

[0030] In the formula, H l+1is the graph convolution feature of the l+1th graph convolution layer, ReLU(·) is the nonlinear activation function, D is the degree matrix, A is the adjacency matrix, and W l is the graph convolution weight of the lth graph convolution layer, is the learning behavior bias weight of the lth layer;

[0031] Step S43: construct a bidirectional long short-term memory subnet, specifically, use the stacked autoencoder data and the graph convolution feature as the input of the bidirectional long short-term memory subnet, and construct a standard bidirectional long short-term memory network, including an input layer, an output layer and a forgetting layer, and construct a bidirectional long short-term memory subnet by calculating the bidirectional hidden state, and then introduce the learning behavior offset weight through the bidirectional improvement mechanism to obtain the bidirectional merge feature, and the calculation formula is:

[0032] ;

[0033] In the formula, h t is a two-way merge feature, is the forward learning behavior bias weight, is the forward hidden state output, is the backward learning behavior bias weight, is the backward hidden state output;

[0034] Step S44: constructing a prediction output layer, specifically, predicting and outputting the performance prediction result through a fully connected layer according to the bidirectional merge feature, and constructing the prediction output layer;

[0035] Step S45: training the performance analysis prediction model, specifically, training the performance analysis prediction model through the stacked feature encoder, the graph feature extraction subnet, the bidirectional long short-term memory subnet and the prediction output layer to obtain the performance analysis prediction model Model CA ;

[0036] Step S46: performance analysis and prediction, specifically using the performance analysis and prediction model Model CA , based on the enhanced student academic performance prediction data and the learning behavior analysis reference data set, performance analysis prediction is performed to obtain a performance analysis prediction data set.

[0037] Further, in step S5, the student academic performance prediction is used to predict the academic performance based on the comprehensive results of the student's learning behavior and the performance prediction, specifically, based on the learning behavior analysis reference data set and the performance analysis prediction data set, data weighted integration and visualization analysis are performed to obtain the student academic performance comprehensive prediction reference data;

[0038] The student academic performance comprehensive prediction reference data specifically includes basic performance prediction data, academic behavior performance improvement trend prediction data, and academic behavior ontology improvement trend prediction data;

[0039] The basic grade prediction number is used to indicate the predicted academic grades of the students in the mid-term and final examinations of the current semester;

[0040] The academic behavior and performance improvement trend prediction data is used to indicate the improvement trend prediction result of the academic performance after the student's learning behavior is improved in the current semester;

[0041] The academic behavior ontology improvement trend prediction data is used to represent the prediction results of the probability of improving the student's learning behavior in the current semester and the improvement trend.

[0042] The present invention provides a student academic performance prediction system based on artificial intelligence, comprising a data collection module, a data preprocessing module, a learning behavior analysis module, a performance analysis and prediction module and a student academic performance prediction module;

[0043] The data collection module is used for data collection, and obtains the original data set of students' academic performance prediction through data collection, and sends the original data set of students' academic performance prediction to the data preprocessing module;

[0044] The data preprocessing module is used for data preprocessing, and obtains enhanced student academic performance prediction data through data preprocessing, and sends the enhanced student academic performance prediction data to the learning behavior analysis module;

[0045] The learning behavior analysis module is used for learning behavior analysis, obtaining a learning behavior analysis reference data set through learning behavior analysis, and sending the learning behavior analysis reference data set to the performance analysis prediction module and the student academic performance prediction module;

[0046] The performance analysis and prediction module is used for performance analysis and prediction, obtains a performance analysis and prediction data set through performance analysis and prediction, and sends the performance analysis and prediction data set to the student academic performance prediction module;

[0047] The student academic performance prediction module is used for predicting the student academic performance, and obtains comprehensive prediction reference data of the student academic performance through predicting the student academic performance.

[0048] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0049] (1) In view of the technical problem that the existing intelligent prediction methods for student grades often focus on the simple prediction of grade values, but this prediction method often ignores the students' active learning behavior and various other complex factors in the academic process, this solution creatively adopts a method of predicting students' academic grades by combining learning behavior with academic performance. Through the learning behavior characteristics at multiple levels and dimensions, it deeply explores the potential reasons for the fluctuation and change of students' academic performance, and improves the prediction depth of student grade prediction;

[0050] (2) In the existing learning behavior analysis methods, there are learning behaviors that have both discrete individual characteristics and common learning behavior characteristics. Therefore, it is necessary to consider complex global characteristics and to mine similar characteristics for the same type of learning behaviors. However, the existing technology ignores the technical problem of common characteristics in the process of learning behavior analysis. This solution creatively uses an improved convolutional neural network combined with nearest neighbor clustering to analyze learning behavior. By clustering learning behaviors, it improves the in-depth mining and feature extraction of common characteristics and individual characteristics, and further optimizes the extraction depth of learning behavior characteristics through mixed pooling, which provides help for subsequent academic performance prediction;

[0051] (3) In view of the technical problem that the existing technology lacks a method to further predict student grades by combining the results of learning behavior analysis in the existing grade analysis and prediction, and the specific use of the results of learning behavior analysis is not limited to simple feature input, but should also be used as a related factor for grade change and trend prediction, this scheme creatively adopts a bidirectional long short-term memory neural network that combines stacked feature encoding and graph feature extraction to perform grade analysis and prediction, optimizes feature extraction through multiple feature extraction methods, and corrects and adjusts the prediction results by taking learning behavior as a weight value. Finally, a bidirectional long short-term memory structure is used to improve the processing capability of time series data, thereby realizing the prediction of student academic performance combined with learning behavior analysis, and improving the overall usability and reliability of the method. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A schematic diagram of a flow chart of a method for predicting student academic performance based on artificial intelligence provided by the present invention;

[0053] Figure 2 A schematic diagram of a student academic performance prediction system based on artificial intelligence provided by the present invention;

[0054] Figure 3 This is a schematic diagram of the process of data preprocessing in step S2;

[0055] Figure 4 This is a flow chart of the learning behavior analysis in step S3;

[0056] Figure 5 This is a flow chart of the performance analysis and prediction process in step S4.

[0057] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0058] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0059] In the description of the present invention, it should be understood that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0060] Example 1, see Figure 1 The present invention provides a method for predicting student academic performance based on artificial intelligence, the method comprising the following steps:

[0061] Step S1: data collection;

[0062] Step S2: data preprocessing;

[0063] Step S3: learning behavior analysis;

[0064] Step S4: performance analysis and prediction;

[0065] Step S5: Prediction of students’ academic performance.

[0066] By performing the above operations, in view of the technical problem that in the existing intelligent prediction methods of student grades, the existing intelligent systems often focus on the simple prediction of grade values, but this prediction method often ignores the students' own active learning behavior and various other complex factors in the academic process, this solution creatively adopts the method of predicting students' academic performance by combining learning behavior with academic performance. Through the learning behavior characteristics at multiple levels and dimensions, it deeply explores the potential reasons for the fluctuations and changes in students' academic performance, and improves the prediction depth of student grade prediction.

[0067] Example 2, see Figure 1 and Figure 2 In step S1, the data collection is used to collect the original data set for predicting students' academic performance, specifically, to obtain the original data set for predicting students' academic performance through student academic surveys and teaching system data collection;

[0068] The original data set for predicting student academic performance includes historical performance data, study habit data, class participation data, social activity data, and mental health data;

[0069] The historical performance data specifically includes the student's academic performance in all subjects, test type classification performance and performance change trend data;

[0070] The study habit data specifically includes daily homework completion data, study time allocation survey data, review frequency data, and preview frequency data;

[0071] The class participation data specifically includes attendance rate data, class interaction rate data and class assignment submission data;

[0072] The social activity data specifically includes extracurricular activity participation data and group assignment completion data;

[0073] The mental health status data specifically include psychological stress survey data and anxiety status survey data.

[0074] Example 3, see Figure 1 , Figure 2 and Figure 3 This embodiment is based on the above embodiment. In step S2, the data preprocessing is used to optimize the original data, specifically to perform basic preprocessing and data enhancement operations on the original data set of student academic performance prediction to obtain enhanced student academic performance prediction data, which specifically includes the following steps:

[0075] Step S21: basic preprocessing, specifically, performing data cleaning, data type conversion, and manual feature extraction operations on the original data set of student academic performance prediction in sequence to obtain a preprocessed performance prediction data set;

[0076] Step S22: data enhancement, specifically, sequentially performing data labeling, synthetic minority class oversampling, temporal feature enhancement, and cluster feature enhancement operations on the preprocessed performance prediction data set to obtain an enhanced performance prediction data set;

[0077] Step S23: data integration, specifically, integrating the label data and feature data in the enhanced performance prediction data set, and dividing the data to obtain training set data and test set data;

[0078] Step S24: data comprehensive preprocessing, specifically, performing data comprehensive preprocessing through the data preprocessing, the data enhancement and the data integration to obtain enhanced student academic performance prediction data;

[0079] The enhanced student academic performance prediction data specifically includes an enhanced learning behavior analysis data set and an enhanced performance analysis prediction data set.

[0080] Example 4, see Figure 1 , Figure 2 and Figure 4 This embodiment is based on the above embodiment. In step S3, the learning behavior analysis is used to analyze the learning behavior of students in their academic process and provide a reference for subsequent performance prediction. Specifically, based on the enhanced student academic performance prediction data, an improved convolutional neural network combined with nearest neighbor clustering is used to perform learning behavior analysis to obtain a learning behavior analysis reference data set.

[0081] The improved convolutional neural network combined with nearest neighbor clustering specifically includes a nearest neighbor clustering subnet and an improved convolutional subnet;

[0082] The step of using the improved convolutional neural network combined with nearest neighbor clustering to perform learning behavior analysis to obtain a learning behavior analysis reference data set includes:

[0083] Step S31: construct a nearest neighbor clustering subnetwork to classify students according to their learning behaviors and aggregate similar learning behavior patterns. Specifically, the nearest neighbor clustering is performed through K-means calculation to obtain clustered student behavior data. The calculation formula is:

[0084] ;

[0085] In the formula, J is the K-means clustering objective function, which is used to calculate the clustered student behavior data, K is the total number of cluster centers, k is the cluster center index, N is the total number of cluster samples, n is the cluster sample index, is the nth input data sample corresponding to the kth cluster center, is the mean of the kth cluster center;

[0086] Step S32: constructing an improved convolutional subnet, specifically constructing a multi-layer convolutional layer, and using a nonlinear activation function and an improved comprehensive pooling operation to perform feature processing, and obtaining convolution feature data based on the clustered student behavior data;

[0087] The improved comprehensive pooling operation specifically refers to an improved hybrid threshold pooling operation, and the calculation formula is:

[0088] ;

[0089] In the formula, FMTA is the convolution feature data after the improved comprehensive pooling operation, MAXPOOL(·) is the maximum pooling operation function, Y is the initial convolution feature, which is used to represent the original features extracted from the clustered student behavior data by the multi-layer convolution layer, T is the mixing threshold, which is used to determine the specific pooling strategy to be executed, and the specific value is 0.7, AVGPOOL(·) is the average pooling operation function;

[0090] Step S33: training the learning behavior analysis model, specifically, training the learning behavior analysis model through the nearest neighbor clustering subnet and the improved convolution subnet, and using the cross entropy loss function as the model loss function to obtain the learning behavior analysis model Model BA ;

[0091] Step S34: learning behavior analysis, specifically using the learning behavior analysis model Model BA , based on the enhanced student academic performance prediction data, learning behavior analysis is performed to obtain a learning behavior analysis reference data set.

[0092] By performing the above operations, in the existing learning behavior analysis methods, there are learning behaviors that have both discrete individual characteristics and common learning behavior characteristics. Therefore, it is necessary to consider complex global characteristics and to mine similarity characteristics for the same type of learning behaviors. However, the existing technology ignores the technical problem of common characteristics in the process of learning behavior analysis. This solution creatively uses an improved convolutional neural network combined with nearest neighbor clustering to analyze learning behaviors. By clustering learning behaviors, it improves the in-depth mining and feature extraction of common characteristics and individual characteristics, and further optimizes the extraction depth of learning behavior characteristics through mixed pooling, which provides assistance for subsequent academic performance prediction.

[0093] Example 5, see Figure 1 , Figure 2 and Figure 5 This embodiment is based on the above embodiment. In step S4, the performance analysis prediction is used to perform performance analysis prediction in combination with the results of student behavior analysis. Specifically, based on the enhanced student academic performance prediction data and the learning behavior analysis reference data set, a bidirectional long short-term memory neural network combining stacked feature coding and graph feature extraction is used to perform performance analysis prediction to obtain a performance analysis prediction data set;

[0094] The bidirectional long short-term memory neural network combining stacked feature encoding and graph feature extraction specifically includes a stacked feature encoder, a graph feature extraction subnet, a bidirectional long short-term memory subnet and a prediction output layer;

[0095] The step of using a bidirectional long short-term memory neural network combining stacked feature encoding and graph feature extraction to perform performance analysis and prediction to obtain a performance analysis and prediction data set includes:

[0096] Step S41: constructing a stacked feature encoder, specifically constructing a multi-layer autoencoder for feature learning, and extracting the enhanced student academic performance prediction data and the learning behavior analysis reference data set as feature data and performing data integration to obtain stacked autoencoder data;

[0097] Step S42: construct a graph feature extraction subnet, specifically, based on the stacked autoencoder data, construct a graph convolutional neural network as a graph feature extraction subnet, and integrate the learning behavior analysis reference data set to generate a learning behavior offset weight, and generate a graph convolution feature. The calculation formula is:

[0098] ;

[0099] In the formula, H l+1 is the graph convolution feature of the l+1th graph convolution layer, ReLU(·) is the nonlinear activation function, D is the degree matrix, A is the adjacency matrix, and W l is the graph convolution weight of the lth graph convolution layer, is the learning behavior bias weight of the lth layer;

[0100] Step S43: construct a bidirectional long short-term memory subnet, specifically, use the stacked autoencoder data and the graph convolution feature as the input of the bidirectional long short-term memory subnet, and construct a standard bidirectional long short-term memory network, including an input layer, an output layer and a forgetting layer, and construct a bidirectional long short-term memory subnet by calculating the bidirectional hidden state, and then introduce the learning behavior offset weight through the bidirectional improvement mechanism to obtain the bidirectional merge feature, and the calculation formula is:

[0101] ;

[0102] In the formula, h t is a two-way merge feature, is the forward learning behavior bias weight, is the forward hidden state output, is the backward learning behavior bias weight, is the backward hidden state output;

[0103] Step S44: constructing a prediction output layer, specifically, predicting and outputting the performance prediction result through a fully connected layer according to the bidirectional merge feature, and constructing the prediction output layer;

[0104] Step S45: training the performance analysis prediction model, specifically, training the performance analysis prediction model through the stacked feature encoder, the graph feature extraction subnet, the bidirectional long short-term memory subnet and the prediction output layer to obtain the performance analysis prediction model Model CA ;

[0105] Step S46: performance analysis and prediction, specifically using the performance analysis and prediction model Model CA , based on the enhanced student academic performance prediction data and the learning behavior analysis reference data set, performance analysis prediction is performed to obtain a performance analysis prediction data set.

[0106] By executing the above operations, in view of the technical problem that in the existing grade analysis and prediction, the existing technology lacks a method to further predict students' grades by combining the results of learning behavior analysis, and the specific use of the results of learning behavior analysis is not limited to simple feature input, but should also be used as a related factor for grade change and trend prediction, this solution creatively adopts a bidirectional long short-term memory neural network that combines stacked feature encoding and graph feature extraction to perform grade analysis and prediction, optimizes feature extraction through multiple feature extraction methods, and corrects and adjusts the prediction results by taking learning behavior as a weight value. Finally, a bidirectional long short-term memory structure is used to improve the processing capability of time series data, thereby realizing the prediction of students' academic performance combined with learning behavior analysis, and improving the overall usability and reliability of the method.

[0107] Example 6, see Figure 1 and Figure 2 , this embodiment is based on the above embodiment. In step S5, the student academic performance prediction is used to predict the academic performance based on the comprehensive results of the student's learning behavior and the performance prediction. Specifically, based on the learning behavior analysis reference data set and the performance analysis prediction data set, data weighted integration and visualization analysis are performed to obtain the student academic performance comprehensive prediction reference data;

[0108] The student academic performance comprehensive prediction reference data specifically includes basic performance prediction data, academic behavior performance improvement trend prediction data, and academic behavior ontology improvement trend prediction data;

[0109] The basic grade prediction number is used to indicate the predicted academic grades of the students in the mid-term and final examinations of the current semester;

[0110] The academic behavior and performance improvement trend prediction data is used to indicate the improvement trend prediction result of the academic performance after the student's learning behavior is improved in the current semester;

[0111] The academic behavior ontology improvement trend prediction data is used to represent the prediction results of the probability of improving the student's learning behavior in the current semester and the improvement trend.

[0112] Embodiment 7, see Figure 1 and Figure 2 , this embodiment is based on the above embodiment, and the present invention provides a student academic performance prediction system based on artificial intelligence, including a data collection module, a data preprocessing module, a learning behavior analysis module, a performance analysis and prediction module and a student academic performance prediction module;

[0113] The data collection module is used for data collection, and obtains the original data set of students' academic performance prediction through data collection, and sends the original data set of students' academic performance prediction to the data preprocessing module;

[0114] The data preprocessing module is used for data preprocessing, and obtains enhanced student academic performance prediction data through data preprocessing, and sends the enhanced student academic performance prediction data to the learning behavior analysis module;

[0115] The learning behavior analysis module is used for learning behavior analysis, obtaining a learning behavior analysis reference data set through learning behavior analysis, and sending the learning behavior analysis reference data set to the performance analysis prediction module and the student academic performance prediction module;

[0116] The performance analysis and prediction module is used for performance analysis and prediction, obtains a performance analysis and prediction data set through performance analysis and prediction, and sends the performance analysis and prediction data set to the student academic performance prediction module;

[0117] The student academic performance prediction module is used for predicting the student academic performance, and obtains comprehensive prediction reference data of the student academic performance through predicting the student academic performance.

[0118] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0119] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0120] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A method for predicting student academic performance based on artificial intelligence, characterized in that: The method comprises the following steps: Step S1: Data collection, used to collect the original data set for student academic performance prediction, specifically, obtaining the original data set for student academic performance prediction through student academic survey and teaching system data collection; The original data set for predicting student academic performance includes historical performance data, study habit data, class participation data, social activity data, and mental health data; Step S2: data preprocessing, which is used to optimize the original data, specifically, to perform basic preprocessing and data enhancement operations on the original data set of student academic performance prediction to obtain enhanced student academic performance prediction data; The enhanced student academic performance prediction data specifically includes an enhanced learning behavior analysis data set and an enhanced performance analysis prediction data set; Step S3: learning behavior analysis, which is used to analyze the learning behavior of students in their academic process and provide a reference for subsequent performance prediction. Specifically, based on the enhanced student academic performance prediction data, an improved convolutional neural network combined with nearest neighbor clustering is used to perform learning behavior analysis to obtain a learning behavior analysis reference data set; The improved convolutional neural network combined with nearest neighbor clustering specifically includes a nearest neighbor clustering subnet and an improved convolutional subnet; The step of using the improved convolutional neural network combined with nearest neighbor clustering to perform learning behavior analysis to obtain a learning behavior analysis reference data set includes: Step S31: construct a nearest neighbor clustering subnetwork to classify students according to their learning behaviors and aggregate similar learning behavior patterns. Specifically, the nearest neighbor clustering is performed through K-means calculation to obtain clustered student behavior data. The calculation formula is: In the formula, J is the K-means clustering objective function, which is used to calculate the clustered student behavior data, K is the total number of cluster centers, k is the cluster center index, N is the total number of cluster samples, n is the cluster sample index, is the nth input data sample corresponding to the kth cluster center, μ k is the mean of the kth cluster center; Step S32: constructing an improved convolutional subnet, specifically constructing a multi-layer convolutional layer, and using a nonlinear activation function and an improved comprehensive pooling operation to perform feature processing, and obtaining convolution feature data based on the clustered student behavior data; The improved comprehensive pooling operation specifically refers to an improved hybrid threshold pooling operation, and the calculation formula is: In the formula, F MTA is the convolution feature data after the improved comprehensive pooling operation, MAXPOOL(·) is the maximum pooling operation function, Y is the initial convolution feature, which is used to represent the original features extracted from the clustered student behavior data by the multi-layer convolution layer, T is the mixing threshold, which is used to determine the specific pooling strategy to be executed, and the specific value is 0.7, AVGPOOL(·) is the average pooling operation function; Step S33: training the learning behavior analysis model, specifically, training the learning behavior analysis model through the nearest neighbor clustering subnet and the improved convolution subnet, and using the cross entropy loss function as the model loss function to obtain the learning behavior analysis model Model BA ; Step S34: learning behavior analysis, specifically using the learning behavior analysis model Model BA , performing learning behavior analysis based on the enhanced student academic performance prediction data to obtain a learning behavior analysis reference data set; Step S4: performance analysis and prediction, which is used to perform performance analysis and prediction in combination with the results of student behavior analysis, specifically, based on the enhanced student academic performance prediction data and the learning behavior analysis reference data set, a bidirectional long short-term memory neural network combining stacked feature coding and graph feature extraction is used to perform performance analysis and prediction to obtain a performance analysis prediction data set; The bidirectional long short-term memory neural network combining stacked feature encoding and graph feature extraction specifically includes a stacked feature encoder, a graph feature extraction subnet, a bidirectional long short-term memory subnet and a prediction output layer; The step of using a bidirectional long short-term memory neural network combining stacked feature encoding and graph feature extraction to perform performance analysis and prediction to obtain a performance analysis and prediction data set includes: Step S41: constructing a stacked feature encoder, specifically constructing a multi-layer autoencoder for feature learning, and extracting the enhanced student academic performance prediction data and the learning behavior analysis reference data set as feature data and performing data integration to obtain stacked autoencoder data; Step S42: construct a graph feature extraction subnet, specifically, based on the stacked autoencoder data, construct a graph convolutional neural network as a graph feature extraction subnet, and integrate the learning behavior analysis reference data set to generate a learning behavior offset weight, and generate a graph convolution feature. The calculation formula is: In the formula, H l+1 is the graph convolution feature of the l+1th graph convolution layer, ReLU(·) is the nonlinear activation function, D is the degree matrix, A is the adjacency matrix, and W l is the graph convolution weight of the lth graph convolution layer, is the learning behavior bias weight of the lth layer; Step S43: construct a bidirectional long short-term memory subnet, specifically, use the stacked autoencoder data and the graph convolution feature as the input of the bidirectional long short-term memory subnet, and construct a standard bidirectional long short-term memory network, including an input layer, an output layer and a forgetting layer, and construct a bidirectional long short-term memory subnet by calculating the bidirectional hidden state, and then introduce the learning behavior offset weight through the bidirectional improvement mechanism to obtain the bidirectional merge feature, and the calculation formula is: In the formula, h t is a two-way merge feature, is the forward learning behavior bias weight, is the forward hidden state output, is the backward learning behavior bias weight, is the backward hidden state output; Step S44: constructing a prediction output layer, specifically, predicting and outputting the performance prediction result through a fully connected layer according to the bidirectional merge feature, and constructing the prediction output layer; Step S45: training the performance analysis prediction model, specifically, training the performance analysis prediction model through the stacked feature encoder, the graph feature extraction subnet, the bidirectional long short-term memory subnet and the prediction output layer to obtain the performance analysis prediction model Model CA ; Step S46: performance analysis and prediction, specifically using the performance analysis and prediction model Model CA , based on the enhanced student academic performance prediction data and the learning behavior analysis reference data set, perform performance analysis and prediction to obtain a performance analysis and prediction data set Step S5: Predicting students' academic performance: performing weighted integration and visual analysis on the learning behavior analysis reference data set and the performance analysis prediction data set to obtain comprehensive prediction reference data on students' academic performance.

2. The method for predicting student academic performance based on artificial intelligence according to claim 1, characterized in that: In step S5, the student academic performance prediction is used to predict the academic performance by integrating the results of the student's learning behavior and the performance prediction, specifically, based on the learning behavior analysis reference data set and the performance analysis prediction data set, data weighted integration and visualization analysis are performed to obtain the student academic performance comprehensive prediction reference data; The student academic performance comprehensive prediction reference data specifically includes basic performance prediction data, academic behavior performance improvement trend prediction data, and academic behavior ontology improvement trend prediction data; The basic grade prediction number is used to indicate the predicted academic grades of the students in the mid-term and final examinations of the current semester; The academic behavior and performance improvement trend prediction data is used to indicate the improvement trend prediction result of the academic performance after the student's learning behavior is improved in the current semester; The academic behavior ontology improvement trend prediction data is used to represent the prediction results of the probability of improving the student's learning behavior in the current semester and the improvement trend.

3. A student academic performance prediction system based on artificial intelligence, used to implement a student academic performance prediction method based on artificial intelligence as described in any one of claims 1-2, characterized in that: It includes data collection module, data preprocessing module, learning behavior analysis module, performance analysis and prediction module and student academic performance prediction module.

4. The student academic performance prediction system based on artificial intelligence according to claim 3 is characterized by: The data collection module is used for data collection, and obtains the original data set of students' academic performance prediction through data collection, and sends the original data set of students' academic performance prediction to the data preprocessing module; The data preprocessing module is used for data preprocessing, and obtains enhanced student academic performance prediction data through data preprocessing, and sends the enhanced student academic performance prediction data to the learning behavior analysis module; The learning behavior analysis module is used for learning behavior analysis, obtaining a learning behavior analysis reference data set through learning behavior analysis, and sending the learning behavior analysis reference data set to the performance analysis prediction module and the student academic performance prediction module; The performance analysis and prediction module is used for performance analysis and prediction, obtains a performance analysis and prediction data set through performance analysis and prediction, and sends the performance analysis and prediction data set to the student academic performance prediction module; The student academic performance prediction module is used for predicting the student academic performance, and obtains comprehensive prediction reference data of the student academic performance through predicting the student academic performance.

Images