CN112116142A - Student score prediction system and method based on deep learning - Google Patents

Student score prediction system and method based on deep learning Download PDF

Info

Publication number
CN112116142A
CN112116142A CN202010961528.0A CN202010961528A CN112116142A CN 112116142 A CN112116142 A CN 112116142A CN 202010961528 A CN202010961528 A CN 202010961528A CN 112116142 A CN112116142 A CN 112116142A
Authority
CN
China
Prior art keywords
module
model
user
data
student
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202010961528.0A
Other languages
Chinese (zh)
Inventor
常亮
郭宗鑫
李龙
古天龙
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guilin University of Electronic Technology
Original Assignee
Guilin University of Electronic Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Guilin University of Electronic Technology filed Critical Guilin University of Electronic Technology
Priority to CN202010961528.0A priority Critical patent/CN112116142A/en
Publication of CN112116142A publication Critical patent/CN112116142A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/21—Design, administration or maintenance of databases
    • G06F16/219—Managing data history or versioning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2455—Query execution
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
    • G06F16/2462—Approximate or statistical queries
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/28—Databases characterised by their database models, e.g. relational or object models
    • G06F16/284—Relational databases
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/20—Education
    • G06Q50/205—Education administration or guidance

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Human Resources & Organizations (AREA)
  • Tourism & Hospitality (AREA)
  • Computational Linguistics (AREA)
  • General Business, Economics & Management (AREA)
  • Evolutionary Computation (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Educational Administration (AREA)
  • Probability & Statistics with Applications (AREA)
  • Health & Medical Sciences (AREA)
  • Marketing (AREA)
  • Educational Technology (AREA)
  • General Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Fuzzy Systems (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Computing Systems (AREA)
  • Evolutionary Biology (AREA)
  • Molecular Biology (AREA)
  • Biophysics (AREA)
  • Operations Research (AREA)
  • Primary Health Care (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Development Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Quality & Reliability (AREA)

Abstract

The invention discloses a student achievement prediction system and method based on deep learning, which relates to the technical field of deep learning, and comprises a data management module and a model operation module, wherein the data management module comprises a user information module and an achievement information module, the user information module is used for realizing the functions of user registration, user login and user information modification, the achievement information module is used for realizing the query of users on achievement information and the modification of achievement information, the model operation module comprises a model training module and a model prediction module, the model training module realizes the initial training of a plurality of times of models, the model prediction module is communicated with the model training module to realize the achievement prediction of the models after the initial training, the deep learning algorithm is utilized, the deep learning can quickly extract the important characteristics of sparse data and process complex nonlinear data, thereby improving the accuracy of the prediction model.

Description

Student score prediction system and method based on deep learning
Technical Field
The invention relates to the technical field of deep learning, in particular to a student score prediction system and a student score prediction method based on deep learning.
Background
With the rise of education data mining, student score prediction becomes one of key tasks of research. The machine learning is used as the core of artificial intelligence, a problem solving method is provided in a data mining task, and great value is embodied.
The student score prediction is a typical data mining task, can evaluate the performance of student courses, provides proper learning guidance for students, and can also provide guidance for teachers to control teaching conditions, improve teaching modes, improve teaching quality and the like. The prior art aims at the problem of student score prediction, and no particularly effective method exists at present.
Disclosure of Invention
The invention aims to provide a student achievement prediction system and method based on deep learning, and aims to solve the problems that in the prior art, a student achievement prediction method is single, the prediction effect is poor, the accuracy is not enough, the result cannot be clearly visualized, and the like.
To achieve the above object, a student score prediction system based on deep learning comprises a data management module and a model operation module,
the data management module comprises a user information module and a score information module, wherein the user information module is used for realizing the functions of user registration, user login and user information modification, and the score information module is used for realizing the inquiry of a user on score information and the modification of the score information;
the model operation module comprises a model training module and a model prediction module, the model training module realizes the initial training of the model for multiple times, and the model prediction module is communicated with the model training module to realize the score prediction of the model after the initial training.
The model operation module further comprises a model evaluation module, and the model evaluation module is communicated with the model training module and used for evaluating the training effect of the model after initial training.
The data management module further comprises a data statistics module, and the data statistics module is communicated with the achievement information module and used for summarizing and analyzing historical achievement data.
The invention also provides a student score prediction method based on deep learning, which is characterized in that,
the user information module classifies users into student users and teacher users;
acquiring the score data of the student user and sending the score data to the score information module;
constructing a data set, and dividing data of different student users into a plurality of data sets;
the teacher user or the student user selects a model in the model training module and inputs the data set into the model training module for training;
and inputting the trained data set into the model prediction module, and analyzing and predicting the score corresponding to the student user by the model prediction module.
In the step of constructing the data set, the data statistics module also obtains the data set production user portrait and correspondingly outputs a statistical graph.
After the step of inputting the data set into the model training module for training, the model evaluation module scores the trained data set and stores the scoring condition.
In the step of analyzing by the model prediction module, the student users can predict their performances, and the teacher user can predict the performances of all the student users.
The student performance prediction system and method based on deep learning comprise a data management module and a model operation module, wherein the model operation module comprises a model training module and a model prediction module, the model training module realizes multiple times of model initial training, the model prediction module is communicated with the model training module to realize performance prediction of a model after initial training, and by utilizing a deep learning algorithm, deep learning can quickly extract important characteristics of sparse data and process complex nonlinear data, so that the accuracy of a prediction model is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without any creative effort.
FIG. 1 is a general block diagram of a student achievement prediction system of the present invention;
FIG. 2 is a block diagram of a data management module according to the present invention;
FIG. 3 is a schematic diagram of a model operation module according to the present invention;
FIG. 4 is a flow chart of the training model acquisition data of the present invention;
FIG. 5 is a flow chart of the training data processing of the present invention;
FIG. 6 is a flow chart of model training of the present invention;
FIG. 7 is a flow chart of model prediction data processing according to the present invention;
FIG. 8 is a flow chart of model prediction according to the present invention;
FIG. 9 is a flow chart of the model evaluation data processing of the present invention;
FIG. 10 is a flow chart of model evaluation according to the present invention.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the drawings are illustrative and are intended to be illustrative of the invention and should not be construed as limiting the invention.
In the description of the present invention, it is to be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," and the like, indicate orientations or positional relationships that are based on the orientations or positional relationships illustrated in the drawings, are used for convenience in describing the invention and to simplify the description, and do not indicate or imply that the referenced devices or elements must have a particular orientation, be constructed in a particular orientation, and be operated, and thus are not to be construed as limiting the invention. Further, in the description of the present invention, "a plurality" means two or more unless specifically defined otherwise.
Referring to fig. 1 to 10, the present invention provides a student achievement prediction system based on deep learning, including a data management module and a model operation module, where the data management module includes a user information module and an achievement information module, the user information module is used to implement functions of user registration, user login and user information modification, and the achievement information module is used to implement user query on achievement information and modification on achievement information; the model operation module comprises a model training module and a model prediction module, the model training module realizes multiple times of initial training on the model, and the model prediction module is communicated with the model training module to realize the performance prediction of the model after initial training. The model operation module further comprises a model evaluation module, and the model evaluation module is communicated with the model training module and used for evaluating the training effect of the model after initial training. The data management module also comprises a data statistics module which is communicated with the achievement information module and used for summarizing and analyzing the achievement data of the previous times.
In the embodiment, the student achievement prediction system constructed based on the student achievement prediction system for deep learning comprises three users, namely a student, a teacher and an administrator. The administrator user is responsible for training and storing the maintenance model according to the student history information provided by the teacher; the student user can inquire the historical information of the student user and predict the future score; the teacher user can query and modify the history of all students and the teacher user can predict the future performance of all students. According to the common functions of different users in the system, the system functions can be divided into a data management module and a model operation module.
In the embodiment of the present invention, the data management module is a basic functional module of the system, and is mainly used for managing and operating personal information of a user or achievement data of a student. The data management module comprises a user information module, a score information module and a data statistics module. The data statistics module can be used by only student users, the score data modification function in the score information module can be used by only teacher users, and on the premise of ensuring privacy and reasonableness, the data management module can provide convenient and visual data management and browsing experience for users in many aspects.
The user information module comprises functions of login, registration and user information modification; the module is mainly used for operating, inquiring or storing the personal information of the user into the database. The user login is an entrance into the system, the user identity and the corresponding user side are established during login, the data set required to be operated by the user is also established, and only after the two items are clear, the form of the system main interface and the operable items can be determined. The score information module comprises score information inquiry and score information modification functions; wherein the query function is available to both student and teacher users and the modification of performance information is available only to teacher users. The score information module is directly connected with the database in a butt joint mode, the data table is directly displayed in the system, and a user can quickly inquire or modify the data table. The data statistics module comprises a score data summarizing and comparing function; the module is a special functional module for student users, mainly integrates and displays the learning condition of students, visualizes the student score data through a reasonable chart, analyzes the personal score information of the students, or compares the personal score information with the score information of other students for reference.
In this embodiment, the model operation module is another functional module of the student achievement prediction system based on deep learning, which is not only convenient for the administrator user to manage the model, but also provides help for the learning of the student user and the teaching work of the teacher user. The model operation module is divided into three modules of model training, model prediction and model evaluation. The system comprises an administrator user available model training and model evaluation module and a student and teacher user available model prediction module. The model training module comprises the functions of model initial training, model continuous training and training visualization. The module is mainly used for training the model and visualizing the training result, so that an administrator can clearly observe the real-time training condition of the model, and meanwhile, necessary preparation is made for starting a prediction module of the model. After the model training module finishes executing, not only the model file is saved, but also the training data volume used by the model training is saved, and a basis is provided for whether the model can be continuously trained or not.
Wherein the model prediction module comprises prediction and visualization functions. After the administrator user generates the model, the student or teacher user can predict the student score through the trained model. The module is the most core function of the system and is the functional module which most embodies the value of the system. The model evaluation module comprises evaluation and visualization functions; compared with the work of model training in the model training module, the model evaluation is different in that the evaluation algorithm is used for screening and storing the optimal model for the trained model.
The invention also provides a student score prediction method based on deep learning, which comprises the steps of obtaining user data and sending the user data to the user information module, wherein the user information module classifies users into student users and teacher users; acquiring the score data of the student user and sending the score data to the score information module; constructing a data set, and dividing data of different student users into a plurality of data sets; the teacher user or the student user selects a model in the model training module, and inputs the data set into the model training module for training; and inputting the trained data set into the model prediction module, analyzing by the model prediction module, and predicting the score corresponding to the student user. The data statistics module also obtains the data set to generate a user portrait and correspondingly outputs a statistical graph. And the model evaluation module scores the trained data set and stores the scoring condition. The student users can predict their own achievements, and the teacher user can predict the achievements of all the student users.
In the implementation step, the user data is acquired to realize the functions of the user information module, including registration, and user information modification. The login function is a function which must be executed when a user enters the system, the user identity and the selected data set information are transmitted to a user side main page through a callback mechanism, and a functional button which is available according with the user identity is provided in a main interface; the registration function is called by a registration button of a login interface, an independent window is popped up, a user is created by inputting necessary user information, and when the user is a student identity, a registered account needs to be a school number to use the student score data analysis and score prediction function matched with the account number; the user information modification function is positioned on a main interface of the system, all users can use the user information modification function, the users can inquire and update own personal basic information in the system, but the identity information and the account number are not allowed to be directly modified.
In the implementation step, a achievement information module is realized, wherein the achievement information module comprises achievement information inquiry and achievement information modification functions. According to the user authority, the student user can only inquire the personal score data but can not modify the personal score data, and the teacher user can inquire and modify the score data of all students; the score information modification function is used by a teacher user, and the teacher can directly double-click the data items of the displayed student score data table to modify the data and save the modified contents in the MySQL database in real time.
In the implementation step, the function of a data statistics module is realized, the data statistics module is a characteristic function which can be used by student users, user figures are constructed according to student score data, and different composition strategies including pie charts, line charts, radar charts and the like exist when different data sets are targeted. When the user uses the function for the first time, the system can process and count student data stored in the database and store various charts, the system cannot repeatedly process the data and generate the charts, and the system fluency and user experience are improved.
It can be understood that there are two main categories of student user information for which are shown in tables 1 and 2:
name of field Field code Type of field Remarks for note
Serial number Id int Main key
Number learning Account varchar Indexing
Examination room seat number Is free of varchar Is free of
Examination question 1 Is free of varchar Is free of
Examination question 2 Is free of varchar Is free of
Examination questions three Is free of varchar Is free of
Examination question four Is free of varchar Is free of
Examination questions five Is free of varchar Is free of
Six examination questions Is free of varchar Is free of
Seven examination questions Is free of varchar Is free of
Examination question eight Is free of varchar Is free of
Rolled noodle score Score varchar Is free of
SPOC Unit test 1 Is free of varchar Is free of
SPOC Unit test 2 Is free of varchar Is free of
SPOC Unit test 3 Is free of varchar Is free of
SPOC Unit test 4 Is free of varchar Is free of
SPOC Unit test 5 Is free of varchar Is free of
SPOC Unit test 6 Is free of varchar Is free of
SPOC Unit test 7 Is free of varchar Is free of
SPOC Unit test 8 Is free of varchar Is free of
SPOC Total score Is free of varchar Is free of
SPOC discussion Is free of varchar Is free of
SPOC examination Is free of varchar Is free of
Classroom testing 1 Is free of varchar Is free of
Classroom testing 2 Is free of varchar Is free of
Classroom testing 3 Is free of varchar Is free of
Classroom testing 4 Is free of varchar Is free of
Classroom testing 5 Is free of varchar Is free of
TABLE 1 first class student user information
Figure BDA0002680729040000061
Figure BDA0002680729040000071
TABLE 2 second class student user information
In this implementation step, a model training module is implemented. The model training module has two functions of model training and training visualization, as shown in fig. 4, and acquires data during training. The model training function comprises two training modes of retraining the model and continuing to train the model. And the administrator user retrains or continues to train the function by using the model, and extracts the selected data set information from the MySQL database. As shown in the data processing flow of FIG. 5, the training function of the model extracts, converts, summarizes and batches necessary training features according to the data characteristics for training each algorithm model. The model training module extracts student score data from a database according to the requirements of the model, uses the preprocessed data for model training, and stores necessary information such as model parameters, graph structures and the like for a user to use the model for prediction. The training of the model also needs to detect the data volume of the data set in advance, the data items supported by the training need to contain the final achievements or performance items of the students, if the data item is null, the system judges that the data item is an item to be predicted or a test item, and the data item is not used as training data. When the number of data meeting the training condition is insufficient, the training can not be carried out. When the model training file is saved, the trained data volume information is also saved, and when the model continuous training function is executed, the system can judge whether the execution condition is met according to the saved training data volume information and the condition of the data set, and opens and limits the function of continuous training.
In the specific embodiment of the step, three models of a logistic regression algorithm diagram structure, a support vector machine diagram structure and a deep neural network diagram structure are included.
Wherein, in the structure of the logistic regression algorithm chart, the parameter to be solved is set as omega1，ω2……ωkAnd b, the relation between the sample vector x and the hidden state p is as follows:
P＝L(b+(ω1，ω2……ωk)x)
wherein L is a conversion function, because the student achievement problem that solves belongs to the multi-classification problem, the L function adopts softmax regression method to normalize the gradient of finite item discrete probability distribution to data, namely the result output of linear function is the probability to a plurality of categories, and the probability that sample vector x is classified as ith category is:
Figure BDA0002680729040000072
the adopted logistic regression algorithm uses cross entropy as a loss function, the classification result is set as R, the real result is set as Y, and the cross entropy loss value is as follows:
Loss＝-∑Y·lnR
the target function is converged to the minimum value by using a random gradient descent algorithm, a cross entropy loss function is set to be Q (W, b), the model learning rate is alpha, and the updating mode of solving parameters is as follows:
Figure BDA0002680729040000081
according to the advantages of development by using a Tensorflow framework, the updating of parameters is accelerated by a dynamic programming method by using an efficient back propagation mode.
In the structure of the support vector machine diagram, a Gaussian kernel skill is adopted to solve the nonlinear multi-classification problem, a radial basis function is used to convert the linear nondifferential problem into the linear separable problem of high latitude, sigma > 0 is the bandwidth of the Gaussian kernel, the scaling is set as gama, and the scaling is expressed as:
Figure RE-GDA0002788310670000083
the expression of the kernel function is:
Figure RE-GDA0002788310670000084
wherein xc is the center of the sum function, | | x-xc | | Y2Is the euclidean distance (L2 norm) of vector x and vector xc.
If the real category is y and the learning parameter is b, the dual problem expression is:
Loss＝-∑(∑b-∑(K*||b||2*||y||2)
similar to the logistic regression solution, training of the SVM model is completed by adopting random gradient descent and back propagation.
In the deep neural network graph structure, a six-layer fully-connected network is adopted for feature transformation and learning, a Relu activation function is adopted for solving the problems of gradient explosion and gradient disappearance, a softmax function is also adopted for loss calculation in combination with a cross entropy loss function, and the optimization method uses the adaptive moment to estimate the learning rate of each parameter of the dynamic modification model. The training process of each algorithm is substantially the same, and the main flow of model training is shown in fig. 6. In the process of model training of the system, the system generates a corresponding line graph according to the Loss and the training accuracy rate generated by the model during training, and continuously updates the line graph. The set number of training steps of each 100 times is used as a horizontal axis unit and an updating point of the line graph, and the line graph is generated every time, replaces the line graph generated last time and is displayed on a system main interface. The dynamic line graph of model training allows the user to observe the model training process in real time.
It can be understood that after the model training is finished, the system displays all Loss and training accuracy rate generated by the algorithm model training on the main interface in a summary mode. The visualization of model training gives more intuitive and clear model training experience to an administrator user, and the main parameters of training are displayed on a system interface in real time, so that the model training work is integrated into the daily system use, and the administrator user can conveniently manage the model. Model training can be stopped as long as the administrator finds a problem in the training process. As long as the model file is not generated, other users cannot use the model for prediction, so that the system can be effectively prevented from being crashed, and abnormal results obtained during prediction by the users can also be prevented.
In this embodiment, a model prediction module is implemented. The model prediction module reads the stored model data file, the student users and the teacher users predict the input student performance data, the student performance data input by the users can be processed in a series, the student users can only predict by inputting the school number, as shown in fig. 7, the teacher users can input the school number for prediction and can also directly predict by double clicking the data item, wherein the data processing flow is as follows. As shown in fig. 8, the prediction process is substantially the same for each model. And obtaining a predicted result corresponding to the student result data input by the user according to the model prediction function, and displaying the predicted result on a system main interface in a visualization chart mode. The model prediction module is a core function of the system, needs an administrator user to carefully optimize the model, and also needs a student and a teacher user to provide more student performance data with mining significance, a model algorithm used by the module can be divided into a linear model and a nonlinear model, and can also be divided into a model suitable for small data and a model suitable for large data, so that the system can always obtain a good prediction effect on some algorithm models no matter what data characteristics are based on, and meanwhile, the module adopts a mode of newly establishing a session and recovering a model structure, improves the system fluency by using the high-performance advantage shown by a Tensorflow frame, and brings better performance prediction experience for the user.
In this embodiment, a model evaluation module is implemented. The model evaluation module is used for evaluating the model by an administrator user, evaluating performances in all algorithm model training by adopting a mature evaluation algorithm, and taking the evaluation accuracy as a measurement standard. The evaluation algorithm used by the module is K-fold cross test, the algorithm strategy is to divide a data set into K groups, train by using K-1 group data, evaluate by using 1 group data, and repeat the process until all the group data are evaluated once, namely, K times of training are carried out. The auxiliary parameter of the evaluation model is a kappa coefficient, and when the model is biased to a category with a large quantitative proportion in the unbalanced classification sample data in the process of evaluation at a certain time, the kappa coefficient obtained by calculating the confusion matrix is low. The model training process in the model evaluation module is different from the model training process in that the data processing process of the model evaluation training is different from that of the model evaluation training module.
As shown in fig. 9, the evaluation process of the model is to add evaluation work on the basis of model training, and after the training work is completed each time, the model is evaluated, and the evaluation result is transmitted to the system main interface to be displayed in real time through a chart, as shown in fig. 10, the model evaluation process. In the process of model evaluation, similar to the model training module, the system displays the evaluation result on the system main interface according to the specific situation of each evaluation, and iteratively updates the chart. After the model evaluation is completed, the system collects the information of the evaluation process into a chart, the information comprises a kappa coefficient and evaluation accuracy, and the evaluation of all algorithm models is uniformly compared and displayed in an intuitive chart form. The model evaluation module gives the administrator user more deep model management. For the adjustment of the model, the training and prediction conditions of the model are usually continuously observed, and the model evaluation based on the K-fold cross test solves the problem and can also be used as a reference for classification accuracy according to the kappa coefficient. And the module has obvious evaluation feedback in a longer evaluation process when being used with a visual effect, so that an administrator user can obtain good model evaluation experience. The model evaluation module stores the model with the highest evaluation score after each evaluation for student users and teacher users, and along with the continuous increase of student score data volume, the performance of the optimal model obtained by model evaluation is improved, and better prediction experience is provided for the users.
The invention provides a student achievement prediction system and method based on deep learning, which adopts three algorithms to predict student achievement, wherein the three algorithms comprise a logistic regression analysis algorithm, a support vector machine algorithm and a deep neural network algorithm. The traditional learning achievement prediction system usually adopts a machine learning algorithm to predict the achievement of students. The invention has three different algorithms which can show good performance under the conditions of small data volume and rich data volume. By utilizing a deep learning algorithm, important features of sparse data can be rapidly extracted through deep learning, and complex nonlinear data is processed, so that the accuracy of a prediction model is improved. A visual interface is constructed, so that a user can clearly observe data and experimental results. The score prediction system has a certain guiding function for the learning of students, a guiding function for the teaching adjustment of teachers, and an important reference value for the algorithm optimization work of administrators. The system realizes a perfect student score prediction system, has comprehensive functions, various user roles and great practical value.
While the invention has been described with reference to a preferred embodiment, it will be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the spirit and scope of the invention.

Claims (7)

1. A student score prediction system based on deep learning is characterized in that the system comprises a data management module and a model operation module,
the data management module comprises a user information module and a score information module, wherein the user information module is used for realizing the functions of user registration, user login and user information modification, and the score information module is used for realizing the inquiry of a user on score information and the modification of the score information;
the model operation module comprises a model training module and a model prediction module, wherein the model training module realizes the initial training of multiple models, and the model prediction module is communicated with the model training module to realize the score prediction of the models after the initial training.
2. The deep learning-based student achievement prediction system of claim 1, wherein the model operation module further comprises a model evaluation module, the model evaluation module is in communication with the model training module for evaluating training effects of the model after initial training.
3. The student achievement prediction system based on deep learning of claim 1, wherein the data management module further comprises a data statistics module, the data statistics module is in communication with the achievement information module for summarizing and analyzing historical achievement data.
4. A student score prediction method based on deep learning, which is characterized in that,
the user information module classifies users into student users and teacher users;
acquiring the score data of the student user and sending the score data to the score information module;
constructing a data set, and dividing data of different student users into a plurality of data sets;
the teacher user or the student user selects a model in the model training module and inputs the data set into the model training module for training;
and inputting the trained data set into the model prediction module, and analyzing and predicting the score corresponding to the student user by the model prediction module.
5. The deep learning-based student performance prediction method of claim 4, wherein in the step of constructing a data set, the data statistics module also obtains the data set to produce a user representation corresponding to an output statistical graph.
6. The deep learning-based student performance prediction method of claim 5, wherein after the step of inputting the data set into the model training module for training, the model evaluation module scores the trained data set and saves the scored data set.
7. The deep learning-based student achievement prediction method of claim 6, wherein in the step of analyzing by the model prediction module, the student user can predict his/her achievement, and the teacher user can predict all of the student user's achievements.
CN202010961528.0A 2020-09-14 2020-09-14 Student score prediction system and method based on deep learning Pending CN112116142A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010961528.0A CN112116142A (en) 2020-09-14 2020-09-14 Student score prediction system and method based on deep learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010961528.0A CN112116142A (en) 2020-09-14 2020-09-14 Student score prediction system and method based on deep learning

Publications (1)

Publication Number Publication Date
CN112116142A true CN112116142A (en) 2020-12-22

Family

ID=73802695

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010961528.0A Pending CN112116142A (en) 2020-09-14 2020-09-14 Student score prediction system and method based on deep learning

Country Status (1)

Country Link
CN (1) CN112116142A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114971095A (en) * 2022-08-02 2022-08-30 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) Online education effect prediction method, device, equipment and storage medium

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102521322A (en) * 2011-12-05 2012-06-27 苏州汉清投资管理有限公司 Score query system
CN106157204A (en) * 2016-06-20 2016-11-23 山东师范大学 A kind of Forecasting The Scores method and system based on BP neural network model
CN108182489A (en) * 2017-12-25 2018-06-19 浙江工业大学 Method is recommended in a kind of individualized learning based on on-line study behavioural analysis
CN108320045A (en) * 2017-12-20 2018-07-24 卓智网络科技有限公司 Student performance prediction technique and device
CN109242149A (en) * 2018-08-13 2019-01-18 华中师范大学 A kind of student performance early warning method and system excavated based on educational data

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102521322A (en) * 2011-12-05 2012-06-27 苏州汉清投资管理有限公司 Score query system
CN106157204A (en) * 2016-06-20 2016-11-23 山东师范大学 A kind of Forecasting The Scores method and system based on BP neural network model
CN108320045A (en) * 2017-12-20 2018-07-24 卓智网络科技有限公司 Student performance prediction technique and device
CN108182489A (en) * 2017-12-25 2018-06-19 浙江工业大学 Method is recommended in a kind of individualized learning based on on-line study behavioural analysis
CN109242149A (en) * 2018-08-13 2019-01-18 华中师范大学 A kind of student performance early warning method and system excavated based on educational data

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114971095A (en) * 2022-08-02 2022-08-30 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) Online education effect prediction method, device, equipment and storage medium
CN114971095B (en) * 2022-08-02 2022-11-08 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) Online education effect prediction method, device, equipment and storage medium

Similar Documents

Publication Publication Date Title
CN106779087A (en) A kind of general-purpose machinery learning data analysis platform
CN106447558A (en) guidance of learning method and learning system combining ontology and clustering analysis technology
CN105518683B (en) The system and method assessed for child's feature recognition and development of potential
CN109165799A (en) The class's of walking education course arrangement system based on genetic algorithm
CN117591944A (en) Learning early warning method and system for big data analysis
CN108874861A (en) A kind of teaching big data Visualized Analysis System and method
CN106202377A (en) A kind of online collaborative sort method based on stochastic gradient descent
CN107103419A (en) One kind of groups software development process analogue system and method
CN112116142A (en) Student score prediction system and method based on deep learning
Tian et al. User intention recognition and requirement elicitation method for conversational ai services
He et al. Representation learning of knowledge graph for wireless communication networks
CN116450704A (en) Automatic generation method and generation device of flow model
Asril et al. Prediction of students study period using K-Nearest Neighbor algorithm
Wang [Retracted] Construction and Effect Analysis of College Students’ Physical Education Teaching Mode Based on Data Mining Algorithm
CN110189236A (en) Alarming system method based on big data
CN114580912A (en) Knowledge distance-based item type learning grouping method and device
Chen et al. Learning disability early warning system based on classification algorithm
Jiang et al. [Retracted] Employment Recommendation for Education Talents Based on Big Data Precision Technology
Yi Research on intelligent evaluation of English diagnosis system based on fuzzy K-means clustering
Ma et al. Study on Predicting University Student Performance Based on Course Correlation
Liu et al. Design of Intelligent Test-composition System for English Test based on RBF Algorithm
Zhang et al. Research on Teaching Quality Evaluation and Group Characteristics of Blended Courses based on Data Mining Technology
Qi Simulation System of College Physical Education Teaching Based on Decision Tree Algorithm and Data Mining
Jia et al. Employment Situation Analysis of College Students Based on Hidden Markov Model
Zhang et al. Auxiliary Teaching Design for Knowledge Points Labeling by Education Data Analysis

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication

Application publication date: 20201222

RJ01 Rejection of invention patent application after publication