Student management method and system based on big data
By optimizing student management methods through probabilistic clustering and random walk algorithms, the problems of adaptive error and improper parameter design in the clustering process are solved, and efficient clustering and management of student behavior data are achieved.
Patent Information
- Application Number
- CN202410921457.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-10
- Publication Date
- 2024-11-05
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing student management methods cannot adapt to different data distributions and clustering structures during clustering. The clustering process has poor adaptability, and improper parameter design leads to insufficient search capabilities and failure to converge quickly.
A probabilistic clustering method is adopted, which uses local density to calculate membership values to allocate data points. A random walk algorithm is introduced to generate transition probabilities and similarity matrices. Local optimization is achieved through the design of the first and second moves. The search efficiency and accuracy are improved by combining adaptive step size adjustment and random strategy.
It improves the accuracy and robustness of clustering results, achieves fast convergence and avoids local optima, and enhances the algorithm's adaptability and search capability.
Smart Images

Figure CN118468069B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of student management, and in particular to a student management method and system based on big data. Background Art
[0002] Student management methods use big data technology and data mining algorithms to collect, process, and analyze students' historical and real-time behavior data, thereby achieving effective management and decision support for student behavior. However, general student management methods have problems such as being unable to adapt to different data distributions and clustering structures when clustering student data, and poor self-adaptation in the clustering process; general student management methods have problems such as improper parameter design during clustering, insufficient local search capabilities of search parameter algorithms, and inability to achieve rapid convergence. Summary of the invention
[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a student management method and system based on big data. In view of the fact that general student management methods cannot adapt to different data distributions and clustering structures when clustering student data, and the clustering process has poor adaptability, this scheme is based on probabilistic clustering, and uses the concept of local density to calculate the membership value to assign data points, so that the clustering process is adaptive; a random walk algorithm is introduced to generate a transition probability matrix and a transition similarity matrix, and the final clustering result is obtained based on the aggregation step, thereby improving the accuracy of the clustering result; in view of the fact that general student management methods have improper parameter design when clustering, the local search ability of the search parameter algorithm is insufficient, and rapid convergence cannot be achieved, this scheme realizes local optimization through the design of primary movement and secondary movement, accelerates convergence based on adaptive adjustment of step size and avoids falling into the local optimal solution; maintains the diversity of the search population through random strategies, improves the robustness of the algorithm, and achieves both search efficiency and accuracy.
[0004] The technical solution adopted by the present invention is as follows: The student management method based on big data provided by the present invention comprises the following steps:
[0005] Step S1: data collection;
[0006] Step S2: data preprocessing;
[0007] Step S3: Probabilistic clustering processing;
[0008] Step S4: Probabilistic clustering optimization;
[0009] Step S5: Student management.
[0010] Furthermore, in step S1, the data collection is to collect student historical behavior data and student real-time behavior data; the student historical behavior data and student real-time behavior data both include student attendance records, student classroom performance evaluations, student academic data, student campus activity participation data, parent feedback data and school rules and regulations violations; the student historical behavior data also includes student behavior status; the student behavior status includes normal student behavior and abnormal student behavior; the student behavior status is used as a data label; the data label does not participate in clustering operations.
[0011] Furthermore, in step S2, the data preprocessing is to perform data cleaning, data conversion and standardization on the collected data; the data cleaning is to process missing values, abnormal values and duplicate values of the collected data; the data conversion is to convert the collected data into a vector form; and the standardization is to standardize the data based on the maximum and minimum normalization method.
[0012] Furthermore, in step S3, the probability clustering process specifically includes the following steps:
[0013] Step S31: data point allocation, initialization of cluster centers; data points are allocated based on the membership values of data objects to obtain basic clusters; the formula used to calculate the membership values is as follows:
[0014] ;
[0015] In the formula, u i1,j1 is the membership of the i1th data point in the j1th cluster; D1 is the total number of feature dimensions of the data, d1 is the feature dimension index; C is the total number of clusters, k1 is the cluster index; x i1d1 is the value of the i1th data point in the d1th feature dimension; v j1d1 and v k1d1 are the values of the j1th cluster and the k1th cluster at the d1th feature dimension respectively; w d1 is the weight of the d1th feature dimension; m(i1,j1) is the local density of the i1th data point relative to the j1th data point, , where ρ i1 and ρ j1 are the number of neighboring data points of the i1th data point and the j1th data point respectively;
[0016] Step S32: Calculate the association matrix, divide the data set into basic partitions in advance, if there is a data point in the same cluster for all basic partitions, the data point is determined to be a sub-cluster; calculate the association matrix based on the sub-cluster, the formula used is as follows:
[0017] ;
[0018] In the formula, CO mc is the association matrix based on sub-clusters; M is the number of basic clusters; c i,j In M basic clusters, subcluster c i and c j The number of occurrences in the same cluster;
[0019] Step S33: Establish a subcluster graph and initialize the subcluster graph, where nodes represent subclusters and edges represent associations between subclusters; an optimization parameter k is preset; select the links between the first k subclusters with the highest association, and delete the remaining links;
[0020] Step S34: Generate a transition probability matrix, pre-set the transition probability parameter T, introduce a random walk algorithm, used to transfer from one sub-cluster node to another sub-cluster node, and the transition probability is proportional to the link weight between the nodes. Generate a transition probability matrix based on the transition probability, and repeat the iteration T times or when the transition probability matrix converges, the generation of the transition probability matrix ends; the formula used to calculate the transition probability is as follows:
[0021] ;
[0022] In the formula, p i,j is from the child cluster node c i to c j The transition probability of n i 、n j and n k They are the sub-cluster nodes c i 、c j and c k The number of nodes; w i,j is the sub-cluster node c i and c j The total number of links between i,k is the sub-cluster node c i and c k The total number of links between
[0023] Step S35: Generate a transfer similarity matrix. For the sub-cluster nodes, generate a probability distribution vector based on the transfer probability matrix. The probability distribution vector of the node represents the transfer probability distribution from the node to all other nodes. Calculate the transfer similarity between vectors based on the probability distribution vector and generate a transfer similarity matrix. The formula used is as follows:
[0024] ;
[0025] Where p and q are the probability distribution vectors of the two sub-cluster nodes; Cos sim is the transfer similarity between two probability distribution vectors;
[0026] Step S36: Aggregation, pre-aggregation parameters of roommates , for each node a in sub-cluster A and each node b in sub-cluster B, take the maximum value of the similarity between nodes as the distance between sub-clusters A and B; based on the distance between sub-clusters, select the two sub-clusters with the smallest distance to aggregate into one cluster; after aggregation, update the transfer similarity matrix; repeat the aggregation until the number of clusters is .
[0027] Furthermore, in step S4, the probability clustering optimization specifically includes the following steps:
[0028] Step S41: Initialization, based on the initial cluster center, optimization parameter k, transition probability parameter T and aggregation parameter Establish a search space and randomly initialize the search population position; cluster the student historical behavior data based on the parameters represented by the search individual position, and select cluster labels based on the labels after clustering converges; use the accuracy of the cluster result data for the cluster label as the individual fitness value;
[0029] Step S42: Initial move 1, the first move position of the bottom 50% of individuals with the lowest fitness value is performed; the formula used is as follows:
[0030] ;
[0031] In the formula, and are the positions of the J-th dimension of the last 50% of the fitness values at the t+1th iteration and the tth iteration respectively; r is a random number in the range of 0 to 1; R is a random number in the range of 1 to 2; c J is the position of the Jth dimension of the population's optimal solution;
[0032] Step S43: Initial move 2, the individuals with the top 50% fitness values are initially moved; the formula used is as follows:
[0033] ;
[0034] In the formula, and are the positions of the J-th dimension of the first individual in the top 50% of fitness values at the t+1th iteration and the tth iteration respectively; w is the moving step length; F I is the individual fitness value; tmax is the maximum number of iterations; is the smoothing term; P I,J is the individual's best historical position;
[0035] Step S44: Secondary movement: perform a second movement on all individuals. The formula used is as follows:
[0036] ;
[0037] In the formula, and They are the position before and after the second move at the t+1th iteration respectively;
[0038] Step S45: Search judgment. A fitness threshold is set in advance. If there is an individual whose fitness value is higher than the fitness threshold, the student's historical behavior data and the student's real-time behavior data are clustered based on the individual position to obtain the clustering result; if the maximum number of iterations is reached, the population position is reinitialized; otherwise, the individuals are re-sorted and the search is iterated.
[0039] Furthermore, in step S5, the student management is based on the clustering results, and a cluster label is selected based on the label of the student's historical behavior data. If the cluster label to which the student's real-time behavior data belongs is abnormal student behavior, the student corresponding to the data is verified and the corresponding teacher is warned to achieve student management.
[0040] The student management system based on big data provided by the present invention comprises a data acquisition module, a data preprocessing module, a probability clustering processing module, a probability clustering optimization module and a student management module;
[0041] The data collection module collects students' historical behavior data and students' real-time behavior data, and sends the data to the data preprocessing module;
[0042] The data preprocessing module performs data cleaning, data conversion and standardization on the collected data, and sends the data to the probability clustering processing module;
[0043] The probability clustering processing module is based on probability clustering, uses the concept of local density to calculate the membership value allocation data points, introduces the random walk algorithm, generates the transition probability matrix and the transition similarity matrix, and obtains the final clustering result based on the aggregation step; and sends the data to the probability clustering optimization module;
[0044] The probability clustering optimization module realizes local optimization through the initial movement and the secondary movement design, designs the movement strategy based on the adaptive adjustment step size and the random strategy; finally realizes the probability clustering optimization; and sends the data to the student management module of the probability clustering processing module;
[0045] The student management module verifies students with abnormal behaviors based on the clustering results and issues warnings to corresponding teachers.
[0046] The beneficial effects achieved by the present invention using the above scheme are as follows:
[0047] (1) In view of the problem that general student management methods cannot adapt to different data distributions and clustering structures when clustering student data, and the clustering process has poor adaptability, this scheme is based on probabilistic clustering and uses the concept of local density to calculate the membership value to assign data points, making the clustering process adaptive. The random walk algorithm is introduced to generate the transition probability matrix and the transition similarity matrix, and the final clustering result is obtained based on the aggregation step to improve the accuracy of the clustering result.
[0048] (2) In view of the problems of improper parameter design during clustering and insufficient local search capability of the search parameter algorithm in general student management methods, which cannot achieve rapid convergence, this scheme realizes local optimization through the design of primary and secondary moves, accelerates convergence based on adaptive adjustment of step size and avoids falling into the local optimal solution; maintains the diversity of the search population through random strategies, improves the robustness of the algorithm, and achieves both search efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] Figure 1 A schematic diagram of the process of the student management method based on big data provided by the present invention;
[0050] Figure 2 A schematic diagram of a student management system based on big data provided by the present invention;
[0051] Figure 3 It is a schematic diagram of the steps of probabilistic clustering processing;
[0052] Figure 4 Schematic flow chart of the steps optimized for probabilistic clustering.
[0053] 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
[0054] 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.
[0055] In the description of the present invention, it is necessary to understand 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 referred system or element must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as a limitation on the present invention.
[0056] Example 1, see Figure 1 The student management method based on big data provided by the present invention comprises the following steps:
[0057] Step S1: Data collection, collecting students' historical behavior data and students' real-time behavior data;
[0058] Step S2: Data preprocessing, data cleaning, data conversion and standardization of the collected data;
[0059] Step S3: Probabilistic clustering processing, based on probabilistic clustering, using the concept of local density to calculate the membership value allocation data points, introducing the random walk algorithm, generating the transition probability matrix and the transition similarity matrix, and obtaining the final clustering result based on the aggregation step;
[0060] Step S4: Probabilistic clustering optimization, achieving local optimization through primary movement and secondary movement design, designing movement strategy based on adaptive adjustment step size and random strategy; finally achieving probabilistic clustering optimization;
[0061] Step S5: Student management.
[0062] Example 2, see Figure 1 This embodiment is based on the above embodiment. In step S1, both the student historical behavior data and the student real-time behavior data include student attendance records, student classroom performance evaluations, student academic data, student campus activity participation data, parent feedback data, and violations of school rules and regulations; the student historical behavior data also includes student behavior status; the student behavior status includes normal student behavior and abnormal student behavior; the student behavior status is used as a data label; the data label does not participate in clustering operations.
[0063] Example 3, see Figure 1 This embodiment is based on the above embodiment. In step S2, data cleaning is to process missing values, abnormal values and duplicate values of the collected data; data conversion is to convert the collected data into vector form; and standardization is to standardize the data based on the maximum and minimum normalization method.
[0064] Example 4, see Figure 1 and Figure 3This embodiment is based on the above embodiment. In step S3, the probability clustering process specifically includes the following steps:
[0065] Step S31: data point allocation, initialization of cluster centers; data points are allocated based on the membership values of data objects to obtain basic clusters; the formula used to calculate the membership values is as follows:
[0066] ;
[0067] In the formula, u i1,j1 is the membership of the i1th data point in the j1th cluster; D1 is the total number of feature dimensions of the data, d1 is the feature dimension index; C is the total number of clusters, k1 is the cluster index; x i1d1 is the value of the i1th data point in the d1th feature dimension; v j1d1 and v k1d1 are the values of the j1th cluster and the k1th cluster at the d1th feature dimension respectively; w d1 is the weight of the d1th feature dimension; m(i1,j1) is the local density of the i1th data point relative to the j1th data point, , where ρ i1 and ρ j1 are the number of neighboring data points of the i1th data point and the j1th data point respectively;
[0068] Step S32: Calculate the association matrix, divide the data set into basic partitions in advance, if there is a data point in the same cluster for all basic partitions, the data point is determined to be a sub-cluster; calculate the association matrix based on the sub-cluster, the formula used is as follows:
[0069] ;
[0070] In the formula, CO mc is the association matrix based on sub-clusters; M is the number of basic clusters; c i,j In M basic clusters, subcluster c i and c j The number of occurrences in the same cluster;
[0071] Step S33: Establish a subcluster graph and initialize the subcluster graph, where nodes represent subclusters and edges represent associations between subclusters; an optimization parameter k is preset; select the links between the first k subclusters with the highest association, and delete the remaining links;
[0072] Step S34: Generate a transition probability matrix, pre-set the transition probability parameter T, introduce a random walk algorithm, used to transfer from one sub-cluster node to another sub-cluster node, and the transition probability is proportional to the link weight between the nodes. Generate a transition probability matrix based on the transition probability, and repeat the iteration T times or when the transition probability matrix converges, the generation of the transition probability matrix ends; the formula used to calculate the transition probability is as follows:
[0073] ;
[0074] In the formula, p i,j is from the child cluster node c i to c j The transition probability of n i 、n j and n k They are the sub-cluster nodes c i 、c j and c k The number of nodes; w i,j is the sub-cluster node c i and c j The total number of links between i,k is the sub-cluster node c i and c k The total number of links between
[0075] Step S35: Generate a transfer similarity matrix. For the sub-cluster nodes, generate a probability distribution vector based on the transfer probability matrix. The probability distribution vector of the node represents the transfer probability distribution from the node to all other nodes. Calculate the transfer similarity between vectors based on the probability distribution vector and generate a transfer similarity matrix. The formula used is as follows:
[0076] ;
[0077] Where p and q are the probability distribution vectors of the two sub-cluster nodes; Cos sim is the transfer similarity between two probability distribution vectors;
[0078] Step S36: Aggregation, pre-aggregation parameters of roommates , for each node a in sub-cluster A and each node b in sub-cluster B, take the maximum value of the similarity between nodes as the distance between sub-clusters A and B; based on the distance between sub-clusters, select the two sub-clusters with the smallest distance to aggregate into one cluster; after aggregation, update the transfer similarity matrix; repeat the aggregation until the number of clusters is .
[0079] By performing the above operations, the general student management method cannot adapt to different data distributions and clustering structures when clustering student data, and the clustering process has poor adaptability. This scheme is based on probabilistic clustering and uses the concept of local density to calculate the membership value to assign data points, so that the clustering process is adaptive; the random walk algorithm is introduced to generate the transition probability matrix and the transfer similarity matrix, and the final clustering result is obtained based on the aggregation step to improve the accuracy of the clustering result.
[0080] Example 5, see Figure 1 and Figure 4 This embodiment is based on the above embodiment. In step S4, the probability clustering optimization specifically includes the following steps:
[0081] Step S41: Initialization, based on the initial cluster center, optimization parameter k, transition probability parameter T and aggregation parameter Establish a search space and randomly initialize the search population position; cluster the student historical behavior data based on the parameters represented by the search individual position, and select cluster labels based on the labels after clustering converges; use the accuracy of the cluster result data for the cluster label as the individual fitness value;
[0082] Step S42: Initial move 1, the first move position of the bottom 50% of individuals with the lowest fitness value is performed; the formula used is as follows:
[0083] ;
[0084] In the formula, and are the positions of the J-th dimension of the last 50% of the fitness values at the t+1th iteration and the tth iteration respectively; r is a random number in the range of 0 to 1; R is a random number in the range of 1 to 2; c J is the position of the Jth dimension of the population's optimal solution;
[0085] Step S43: Initial move 2, the individuals with the top 50% fitness values are initially moved; the formula used is as follows:
[0086] ;
[0087] In the formula, and are the positions of the J-th dimension of the first individual in the top 50% of fitness values at the t+1th iteration and the tth iteration respectively; w is the moving step length; F I is the individual fitness value; tmax is the maximum number of iterations; is the smoothing term; P I,J is the individual's best historical position;
[0088] Step S44: Secondary movement: perform a second movement on all individuals. The formula used is as follows:
[0089] ;
[0090] In the formula, and They are the position before and after the second move at the t+1th iteration respectively;
[0091] Step S45: Search judgment. A fitness threshold is set in advance. If there is an individual whose fitness value is higher than the fitness threshold, the student's historical behavior data and the student's real-time behavior data are clustered based on the individual position to obtain the clustering result; if the maximum number of iterations is reached, the population position is reinitialized; otherwise, the individuals are re-sorted and the search is iterated.
[0092] By performing the above operations, in order to solve the problems of improper parameter design during clustering and insufficient local search capability of search parameter algorithm in general student management methods, which cannot achieve rapid convergence, this scheme realizes local optimization through the design of primary and secondary moves, accelerates convergence based on adaptive adjustment of step size and avoids falling into local optimal solution; maintains the diversity of search population through random strategies and improves the robustness of the algorithm; and achieves both search efficiency and accuracy.
[0093] Example 6, see Figure 1 This embodiment is based on the above embodiment. In step S5, student management is based on the clustering results. Cluster labels are selected based on the labels of students' historical behavior data. If the cluster label to which the student's real-time behavior data belongs is abnormal student behavior, the student corresponding to the data is verified and the corresponding teacher is warned to achieve student management.
[0094] Embodiment 7, see Figure 2 , this embodiment is based on the above embodiment, the student management system based on big data provided by the present invention includes a data acquisition module, a data preprocessing module, a probability clustering processing module, a probability clustering optimization module and a student management module;
[0095] The data collection module collects students' historical behavior data and students' real-time behavior data, and sends the data to the data preprocessing module;
[0096] The data preprocessing module performs data cleaning, data conversion and standardization on the collected data, and sends the data to the probability clustering processing module;
[0097] The probability clustering processing module is based on probability clustering, uses the concept of local density to calculate the membership value allocation data points, introduces the random walk algorithm, generates the transition probability matrix and the transition similarity matrix, and obtains the final clustering result based on the aggregation step; and sends the data to the probability clustering optimization module;
[0098] The probability clustering optimization module realizes local optimization through the initial movement and the secondary movement design, designs the movement strategy based on the adaptive adjustment step size and the random strategy; finally realizes the probability clustering optimization; and sends the data to the student management module of the probability clustering processing module;
[0099] The student management module verifies students with abnormal behaviors based on the clustering results and issues warnings to corresponding teachers.
[0100] 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.
[0101] 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.
[0102] 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 student management method based on big data, characterized by: The method comprises the following steps: Step S1: data collection; Step S2: data preprocessing; Step S3: Probabilistic clustering processing, based on probabilistic clustering, using the concept of local density to calculate the membership value allocation data points, introducing the random walk algorithm, generating the transition probability matrix and the transition similarity matrix, and obtaining the final clustering result based on the aggregation step; Step S4: Probabilistic clustering optimization, achieving local optimization through primary movement and secondary movement design, designing movement strategy based on adaptive adjustment step size and random strategy; finally achieving probabilistic clustering optimization; Step S5: student management; Step S3 includes step S31: data point allocation, initializing cluster centers; allocating data points based on the membership values of data objects to obtain basic clusters; and the formula used to calculate the membership values is as follows: ; In the formula, u i1,j1 is the membership of the i1th data point in the j1th cluster; D1 is the total number of feature dimensions of the data, d1 is the feature dimension index; C is the total number of clusters, k1 is the cluster index; x i1d1 is the value of the i1th data point in the d1th feature dimension; v j1d1 and v k1d1 are the values of the j1th cluster and the k1th cluster at the d1th feature dimension respectively; w d1 is the weight of the d1th feature dimension; m(i1,j1) is the local density of the i1th data point relative to the j1th data point, , where ρ i1 and ρ j1 are the number of neighboring data points of the i1th data point and the j1th data point respectively; In step S3, the probability clustering process specifically includes the following steps: Step S31: data point allocation; Step S32: Calculate the association matrix, divide the data set into basic partitions in advance, if there is a data point in the same cluster for all basic partitions, the data point is determined to be a sub-cluster; calculate the association matrix based on the sub-cluster, the formula used is as follows: ; In the formula, CO mc is the association matrix based on sub-clusters; M is the number of basic clusters; c i,j In M basic clusters, subcluster c i and c j The number of occurrences in the same cluster; Step S33: Establish a subcluster graph and initialize the subcluster graph, where nodes represent subclusters and edges represent associations between subclusters; an optimization parameter k is preset; select the links between the first k subclusters with the highest association, and delete the remaining links; Step S34: Generate a transition probability matrix, pre-set the transition probability parameter T, introduce a random walk algorithm, used to transfer from one sub-cluster node to another sub-cluster node, and the transition probability is proportional to the link weight between the nodes. Generate a transition probability matrix based on the transition probability, and repeat the iteration T times or when the transition probability matrix converges, the generation of the transition probability matrix ends; the formula used to calculate the transition probability is as follows: ; In the formula, p i,j is from the child cluster node c i to c j The transition probability of n i 、n j and n k They are the sub-cluster nodes c i 、c j and c k The number of nodes; w i,j is the sub-cluster node c i and c j The total number of links between i,k is the sub-cluster node c i and c k The total number of links between Step S35: Generate a transfer similarity matrix. For the sub-cluster nodes, generate a probability distribution vector based on the transfer probability matrix. The probability distribution vector of the node represents the transfer probability distribution from the node to all other nodes. Calculate the transfer similarity between vectors based on the probability distribution vector and generate a transfer similarity matrix. The formula used is as follows: ; Where p and q are the probability distribution vectors of the two sub-cluster nodes; Cos sim is the transfer similarity between two probability distribution vectors; Step S36: Aggregation, pre-aggregation parameters of roommates , for each node a in sub-cluster A and each node b in sub-cluster B, take the maximum value of the similarity between nodes as the distance between sub-clusters A and B; based on the distance between sub-clusters, select the two sub-clusters with the smallest distance to aggregate into one cluster; after aggregation, update the transfer similarity matrix; repeat the aggregation until the number of clusters is ; In step S4, the probability clustering optimization specifically includes the following steps: Step S41: Initialization, based on the initial cluster center, optimization parameter k, transition probability parameter T and aggregation parameter Establish a search space and randomly initialize the search population position; cluster the student historical behavior data based on the parameters represented by the search individual position, and select cluster labels based on the labels after clustering converges; use the accuracy of the cluster result data for the cluster label as the individual fitness value; Step S42: Initial move 1, the first move position of the bottom 50% of individuals with the lowest fitness value is performed; the formula used is as follows: ; In the formula, and are the positions of the J-th dimension of the last 50% of the fitness values at the t+1th iteration and the tth iteration respectively; r is a random number in the range of 0 to 1; R is a random number in the range of 1 to 2; c J is the position of the Jth dimension of the population's optimal solution; Step S43: Initial move 2, the individuals with the top 50% fitness values are initially moved; the formula used is as follows: ; In the formula, and are the positions of the J-th dimension of the first individual in the top 50% of fitness values at the t+1th iteration and the tth iteration respectively; w is the moving step length; F I is the individual fitness value; tmax is the maximum number of iterations; is the smoothing term; P I,J is the individual's best historical position; Step S44: Secondary movement: perform a second movement on all individuals. The formula used is as follows: ; In the formula, and They are the position before and after the second move at the t+1th iteration respectively; Step S45: Search judgment, a fitness threshold is set in advance, if there is an individual whose fitness value is higher than the fitness threshold, the student historical behavior data and the student real-time behavior data are clustered based on the individual position to obtain the clustering result; if the maximum number of iterations is reached, the population position is reinitialized; otherwise, the individuals are re-sorted and iteratively searched; In step S1, the data collection is to collect student historical behavior data and student real-time behavior data; the student historical behavior data and student real-time behavior data both include student attendance records, student classroom performance evaluations, student academic data, student campus activity participation data, parent feedback data and school rules and regulations violations; the student historical behavior data also includes student behavior status; the student behavior status includes normal student behavior and abnormal student behavior; the student behavior status is used as a data label; the data label does not participate in clustering operations; In step S5, the student management is to select cluster labels based on the labels of the student's historical behavior data according to the clustering results. If the cluster label to which the student's real-time behavior data belongs is abnormal student behavior, the student corresponding to the data is verified and the corresponding teacher is warned to achieve student management; In step S2, the data preprocessing is to perform data cleaning, data conversion and standardization on the collected data; the data cleaning is to process missing values, abnormal values and duplicate values of the collected data; the data conversion is to convert the collected data into a vector form; the standardization is to standardize the data based on the maximum and minimum normalization method.
2. A student management system based on big data, used to implement the student management method based on big data as claimed in claim 1, characterized in that: It includes data collection module, data preprocessing module, probability clustering processing module, probability clustering optimization module and student management module; The data collection module collects students' historical behavior data and students' real-time behavior data, and sends the data to the data preprocessing module; The data preprocessing module performs data cleaning, data conversion and standardization on the collected data, and sends the data to the probability clustering processing module; The probability clustering processing module is based on probability clustering, uses the concept of local density to calculate the membership value allocation data points, introduces the random walk algorithm, generates the transition probability matrix and the transition similarity matrix, and obtains the final clustering result based on the aggregation step; And send the data to the probability clustering optimization module; The probability clustering optimization module realizes local optimization through the initial movement and the secondary movement design, designs the movement strategy based on the adaptive adjustment step size and the random strategy; finally realizes the probability clustering optimization; and sends the data to the student management module of the probability clustering processing module; The student management module verifies students with abnormal behaviors based on the clustering results and issues warnings to corresponding teachers.
