An AI-based teaching quality evaluation system

Abstract

The present invention relates to the field of artificial intelligence, and in particular to an AI-based teaching quality evaluation system. Traditional teaching quality evaluation methods often rely on teachers' subjective judgments and students' test scores, and are unable to comprehensively and objectively evaluate teaching effectiveness. The present invention provides such an AI-based teaching quality evaluation system, which uses a multi-faceted evaluation model to evaluate students, so that the evaluation results are more objective and comprehensive, including: a data collection module, a data preprocessing module, a learning situation analysis model, a first student evaluation model, a second student evaluation model, an evaluation integration module, and a visualization interface. The system uses a deep learning model to clarify the relationship between students' attendance, in-class test scores, and homework completion, thereby further reflecting students' classroom performance, making student evaluation reports more accurate and more objective.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence, and in particular to an AI-based teaching quality evaluation system. Background Art

[0002] With the continuous development and popularization of artificial intelligence technology, the field of education has also begun to gradually introduce AI technology to improve teaching quality. However, traditional teaching quality evaluation methods often rely on teachers' subjective judgment and students' test scores, and cannot comprehensively and objectively evaluate teaching effectiveness. In addition, the existing teaching quality evaluation system considers too single aspects, resulting in an evaluation that is not objective and comprehensive enough. In addition, the amount of data is huge and difficult to process. It is difficult for traditional teaching evaluation systems to quickly mine the focus of student evaluation from a large amount of data, which leads to teachers, students, and parents being unable to make valuable suggestions to students based on student evaluation reports to promote their future development. Therefore, there is an urgent need for an AI-based teaching quality evaluation system that can evaluate teaching quality in real time based on students' learning conditions and the characteristics of teaching content, and provide targeted feedback and improvement suggestions for teachers and students. Summary of the invention

[0003] (1) Technical issues to be resolved

[0004] In order to overcome the shortcomings of existing evaluation systems or mechanisms that are not comprehensive and objective, the technical problem to be solved by the present invention is to provide an AI-based teaching quality evaluation system.

[0005] (2) Technical solution

[0006] In order to solve the above technical problems, the present invention provides such an AI-based teaching quality evaluation system, which uses a multi-faceted evaluation model to evaluate students, making the evaluation results more objective and comprehensive. The system includes:

[0007] Data collection module: data collection, collecting student data;

[0008] Data preprocessing module: data preprocessing, data cleaning, format conversion, and normalization;

[0009] Learning situation analysis model: The learning situation analysis model is established, the pre-processed data is input into the model, the future performance is predicted and analyzed, and the evaluation results are output;

[0010] First student evaluation model: integrate the student attendance date, the student's in-class test score and the student's homework completion status and input them into the first student evaluation model to output the first student evaluation report;

[0011] Second student evaluation model: collect student information by questionnaire, and use it as data to input into the second student evaluation model, and output the second student evaluation report;

[0012] Evaluation integration module: Use the entropy weight method to determine the weights of the learning situation analysis model, the first student evaluation model and the second student evaluation model, and integrate the evaluation results of the learning situation analysis model, the first student evaluation model and the second student evaluation model with the evaluation report according to the weights to form an overall evaluation of the students;

[0013] Visual interface: Use the Python Flask framework to build a visual interface, connect the learning situation analysis model, the first student evaluation model and the second student evaluation model, summarize the generated evaluation report and display relevant data and charts, and save the evaluation page if necessary to generate a PDF file.

[0014] Preferably, the data collection module obtains student evaluations and information from multiple aspects, including:

[0015] S201 obtains students’ previous test scores and rankings through the school management system;

[0016] S202 Students fill in the questionnaire set by the evaluation system;

[0017] S203 queries the student's daily performance record filled in by the teacher.

[0018] Preferably, the data preprocessing module cleans, converts the format, and normalizes the collected data, including:

[0019] S301 converts all the collected test scores of students into percentage system for the convenience of subsequent analysis;

[0020] S302 converts the data in the questionnaire into a csv format file;

[0021] S303 extracts key information from the student's attendance and the student's daily performance records filled in by the teacher to obtain the data required by the model;

[0022] S304 data cleaning is to delete the data irrelevant to the prediction in the student daily performance records filled in by the teacher through regular matching.

[0023] Preferably, the learning situation analysis model predicts and analyzes future test scores and trends based on previous test scores, including:

[0024] S401 extracts features, extracts relevant features based on previous test scores, including the lowest score, highest score, average score, and score fluctuations.

[0025] The score fluctuation is determined by calculating the standard deviation and coefficient of variation:

[0026] ;

[0027] in represents the standard deviation, Indicates the scores of all previous exams. represents the average score, Indicates the number of exams.

[0028] ;

[0029] Where CV is the coefficient of variation, is the standard deviation, Indicates the number of previous exams;

[0030] S402: The learning situation analysis model training is performed using a multiple linear regression model based on the extracted features;

[0031] S403 Evaluation and prediction of the learning situation analysis model: Evaluate the learning situation analysis model through training results and test results using the test set, adjust the learning situation analysis model parameters and optimize the learning situation analysis model structure, and predict future test scores and output evaluation results. The evaluation results are presented in text and charts, and the evaluation results include: Standardized processing values ​​of scores , predicted scores and score analysis.

[0032] Preferably, the multivariate linear regression model in S402 includes:

[0033] S4021 Establish a multiple regression model:

[0034] ;

[0035] in is the explained variable, For all the test scores, is the error term,

[0036] By using the least squares method The estimated value of the student's future performance is then predicted and analyzed. The specific process includes:

[0037] First, collect all the test scores of previous students as N groups of observations to obtain N groups of independent variable and dependent variable data;

[0038] Substitute the expression of the multivariate regression model and treat the error term ε as obeying a normal distribution with a mean of 0;

[0039] Fit the observed values ​​and calculate the model prediction value for each observed value ;

[0040] Calculate observations With the model prediction value The square difference between them is summed up to get the error sum of squares SSR (Sumof Squares Residuals)

[0041] ;

[0042] By minimizing the sum of squared errors SSR, the estimated values ​​of the regression coefficients are obtained;

[0043] Using the matrix form of the least squares method, we can obtain the estimated values ​​of the regression coefficients by solving the normal equations. ;

[0044] Through training on a large number of data sets The specific value of , and then predict and analyze the student's future scores;

[0045] S4022 Model evaluation: Evaluate the multivariate regression model through MSE, RMSE, and R-squared values ​​to further adjust the parameters of the multivariate regression model and optimize the structure of the multivariate regression model.

[0046] ;

[0047] ;

[0048] ;

[0049] Where m is the number of samples, is the actual value, is the predicted value, is the average of the actual values.

[0050] Preferably, the first student evaluation model collects the student attendance date, the student in-class test score and the student homework completion status in the student daily performance record filled out by the teacher, and performs an overall evaluation of the student's daily performance in class through a deep learning algorithm, including:

[0051] S601 Data quantification: Quantify and classify the student's daily performance record filled in by the teacher after the cleaning in S304 to obtain the student's attendance date, the student's in-class test score and the student's homework completion status, wherein the student's attendance date and the student's in-class test score are expressed in numbers, and the student's homework completion status is divided into four levels and expressed in four special scores; S602 Curve construction: The student's attendance date, the student's in-class test score and the student's homework completion status are respectively generated in chronological order to generate the student's attendance curve, the student's in-class test score change curve and the student's homework completion curve;

[0052] S603 generates a functional relationship: establishes a corresponding functional relationship between the student's attendance curve, the student's in-class test score change curve, and the student's homework completion curve through a deep learning model and obtains a correlation coefficient, thereby establishing a correlation function between the student's attendance curve, the student's in-class test score change curve, and the student's homework completion curve, and determines the relationship between the student's attendance date, the student's in-class test score, and the student's homework completion status through the correlation function;

[0053] S604. Evaluation report generation: for students’ classroom performance, text generation and visualization are used to display relevant data and charts of the evaluation results. The first evaluation report contains the standardized processing value of the score. and evaluation results.

[0054] Preferably, the use of a deep learning model to establish a corresponding functional relationship and obtain a correlation coefficient is to confirm the student's attendance rate, in-class learning situation, and after-class review situation by clarifying the relationship between the student's attendance date, the student's in-class test score, and the student's homework completion situation, and determine the impact of the student's attendance rate, in-class learning situation, and after-class review situation on each other, thereby quantifying the functional relationship between the student's classroom performance and the student's attendance rate, in-class learning situation, and after-class review situation.

[0055] Preferably, the formula for quantifying students' classroom performance generated by the deep learning model is:

[0056] ;

[0057] in, , , is the weight coefficient, Represents the student's classroom performance, Represents the total number of days, Represents different dates in the total number of days, Indicates whether the student is present on different dates. Represents the scores of students' in-class tests on different dates. The score representing the completion of the student's assignment.

[0058] Preferably, the second student evaluation model obtains student information by setting a questionnaire and makes an objective evaluation of the students according to the questionnaire filling results, including:

[0059] S901 Formulate questionnaire questions and evaluation objectives based on the survey content and purpose;

[0060] S902 pre-processes the collected data to ensure the accuracy and completeness of the data;

[0061] S903 analyzes the preprocessed data and extracts information related to the evaluation index;

[0062] S904 constructs a model based on the statistical analysis results. The second student evaluation model is constructed using a neural network algorithm and appropriate features are selected for modeling;

[0063] S905 evaluates and optimizes the second student evaluation model, where the evaluation indicators include precision, recall, and F1 score, and optimizes and adjusts the second student evaluation model according to the evaluation results;

[0064] Recall

[0065] ;

[0066] TP stands for True Positive, which is the number of samples correctly predicted as abnormal; FN stands for False Negative, which is the number of samples incorrectly predicted as normal.

[0067] F1 Score

[0068] ;

[0069] in represents abnormal samples, Represents the overall sample, Precision represents the precision, and Recall represents the recall rate;

[0070] S906 generates an evaluation report, using text generation and visualization to display relevant data and charts of the evaluation results. The evaluation report includes the standardized processing value of the score and evaluation results.

[0071] Preferably, the S6 evaluation integration module uses the entropy weight method to standardize the learning situation analysis model. , the first student evaluation model , the standardized value of the results of the second student evaluation model The scores of the three models are standardized and integrated into a more accurate and objective evaluation report for students. , and obtain the relative index value of the i-th index by range scaling , the information entropy corresponding to each indicator i for:

[0072] ;

[0073] Final score calculation :

[0074] ;

[0075] in, To calculate the final score, is information entropy; Used to indicate the student's final score.

[0076] (3) Beneficial effects

[0077] The invention uses multiple student evaluation indicators to evaluate the quality of students, including filling out questionnaires, retrieving students' previous grades from the school management platform, and having teachers fill out student evaluations and classroom performance, so as to have a more comprehensive understanding of the student situation and make the evaluation results more accurate. The system uses a deep learning model, so that the model ultimately clarifies the relationship between students' attendance, in-class test scores, and homework completion, thereby further reflecting students' classroom performance and making the evaluation report on students more accurate and more objective. Finally, the entropy weight method is used to scientifically and reasonably assign weights to the results generated by the learning situation analysis model, the first student evaluation model, and the second student evaluation model, so that the results achieved are more comprehensive and accurate, and a visual page is set to make the student's quality evaluation report more intuitive and easy to understand. If the user needs it, the page-based quality evaluation report can be generated into a pdf file for user convenience. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 It is a flow chart of the teaching quality evaluation system of the present invention.

[0079] Figure 2 It is a model flow chart of the teaching quality evaluation system of the present invention. DETAILED DESCRIPTION

[0080] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0081] Example 1

[0082] The present invention provides an AI-based teaching quality evaluation system, and the adaptive teaching quality evaluation system includes a data collection module, a data preprocessing module, a learning situation analysis model, a first student evaluation model, a second student evaluation model, an evaluation integration module and a visualization interface.

[0083] The data collection module obtains data by retrieving student scores from the school management platform, student evaluations filled out by teachers, classroom performance, and questionnaires set up by the system. The data preprocessing module cleans and standardizes the collected data to improve the accuracy and availability of the data. The learning situation analysis model simulates the collected test scores through a multivariate linear regression model to predict future test scores. The first student evaluation model analyzes students' classroom performance and uses a deep learning model to analyze it, and finally obtains students' classroom performance. The second student evaluation model evaluates the questionnaires answered by students and obtains the results. The evaluation integration module distributes weights and integrates the evaluation reports obtained from the three modules to obtain a more detailed, accurate and objective student evaluation report. The system is equipped with a visual interface to display the three evaluation reports and the comprehensive evaluation report, so that students can understand more intuitively, and provide PDF files when students need them.

[0084] Example 2

[0085] like Figure 1 , Figure 2 As shown, an AI-based teaching quality evaluation system includes:

[0086] Data collection module:

[0087] The teaching quality evaluation system obtains students’ previous test scores and rankings by retrieving data from the school management platform;

[0088] The teaching quality evaluation system obtains students’ evaluation information through the classroom performance of old students;

[0089] The teaching quality evaluation system obtains more detailed information about students by asking them to fill out questionnaires within the system.

[0090] The data preprocessing module processes the collected data to ensure the integrity and accuracy of the acquired data and facilitate the generation of subsequent evaluation reports, including:

[0091] Convert all the collected student test scores into percentages for easy subsequent analysis;

[0092] Convert the data in the questionnaire into a csv format file;

[0093] Extract key information from students’ attendance and daily performance records filled out by teachers to obtain the data needed for the model;

[0094] Data cleaning is to delete the data irrelevant to the prediction in the daily performance records of students filled in by teachers through regular matching. The learning situation analysis model imports the processed information of students' previous test scores into the model to predict and analyze future tests, including:

[0095] Extract features, extract relevant features based on the processed data, including the lowest score, the highest score, the average score, and the score fluctuation. The score fluctuation is determined by calculating the standard deviation and the coefficient of variation.

[0096] ;

[0097] in represents the standard deviation, Indicates the scores of all previous exams. represents the average score, Indicates the number of exams.

[0098] ;

[0099] Where CV is the coefficient of variation, is the standard deviation, Indicates the number of previous exams;

[0100] Model training, using a multivariate linear regression model, is performed based on the extracted features, including:

[0101] Build a multiple regression model:

[0102] ;

[0103] in is the explained variable, For all the test scores, is the error term,

[0104] By using the least squares method The specific process of estimating and analyzing students' future scores includes:

[0105] By using the least squares method The specific process of estimating and analyzing students' future scores includes:

[0106] First, collect all the test scores of previous students as N groups of observations to obtain N groups of independent variable and dependent variable data;

[0107] Substitute the expression of the multivariate regression model and treat the error term ε as obeying a normal distribution with a mean of 0;

[0108] Fit the observed values ​​and calculate the model prediction value for each observed value ;

[0109] Calculate observations With the model prediction value The square difference between them is summed up to get the error sum of squares SSR (Sumof Squares Residuals)

[0110] ;

[0111] By minimizing the sum of squared errors SSR, the estimated values ​​of the regression coefficients are obtained;

[0112] Using the matrix form of the least squares method, we can obtain the estimated values ​​of the regression coefficients by solving the normal equations. ;

[0113] First, convert the expression of the multiple regression model into a matrix form. Assuming there are N observations and k independent variables, the observations and the corresponding independent and dependent variables are expressed in the following matrix form:

[0114] ;

[0115] in, is an N-dimensional column vector representing the observed values ​​of the dependent variable, and X is a The first column is all 1, indicating the intercept, and the remaining columns are the observed values ​​of the independent variables. β is a dimensional column vector, representing the regression coefficients of the respective variables, and ε is an n-dimensional column vector, representing the error term.

[0116] Solve the normal equations. Substitute the above matrix form into the minimization objective function to obtain the error sum of squares SSR, and then take the derivative of β, set the derivative to 0, and obtain the normal equations:

[0117] ;

[0118] Among them, ^T represents the transpose operation of the matrix.

[0119] Solve the normal equations. Convert the normal equations into matrix form and write them as:

[0120] ;

[0121] at this time, is a The matrix of is a dimensional column vector, and by solving the above linear equations, we can get the estimated value β of the regression coefficient.

[0122] Model evaluation: The multivariate regression model is evaluated through MSE, RMSE, and R-squared values ​​to further adjust the model parameters and optimize the multivariate regression model structure.

[0123] ;

[0124] ;

[0125] ;

[0126] Where m is the number of samples, is the actual value, is the predicted value, is the average of the actual values.

[0127] The multivariate regression model is evaluated through training results and test results using the test set, model parameters are adjusted and the multivariate regression model structure is optimized, and future test scores are predicted.

[0128] Input the previous test scores and rankings of the students to be evaluated, and the multivariate regression model will output the evaluation results, which will be presented in the form of text and charts. The evaluation results include: the standardized value of the score , predicted scores and score analysis.

[0129] The first student evaluation model mainly evaluates students through daily performance records filled out by teachers, including:

[0130] The daily performance records of students filled in by the teacher after cleaning in S304 are quantified and classified to obtain the attendance date, the scores of the in-class tests and the completion of the homework of the students, wherein the attendance date and the scores of the in-class tests of the students are represented by numbers, and the completion of the homework of the students is divided into four levels and represented by four special scores; constructing curves: the attendance date, the scores of the in-class tests of the students and the completion of the homework of the students are respectively generated according to the time sequence to form the attendance curve of the students, the curve of the change of the scores of the in-class tests of the students and the curve of the completion of the homework of the students;

[0131] The corresponding functional relationship between the student's attendance curve, the student's in-class test score change curve and the student's homework completion curve is established through a deep learning model, and the correlation coefficient is obtained, so as to establish a correlation function between the student's attendance curve, the student's in-class test score change curve and the student's homework completion curve, and determine the relationship between the student's attendance date, the student's in-class test score and the student's homework completion status through the correlation function.

[0132] The purpose of using a deep learning model to establish a corresponding functional relationship and obtain a correlation coefficient is to: confirm the student's attendance rate, in-class learning situation, and after-class review situation by clarifying the relationship between the student's attendance date, the student's in-class test score, and the student's homework completion situation, and determine the impact of the student's attendance rate, in-class learning situation, and after-class review situation on each other, thereby quantifying the functional relationship between the student's classroom performance and the student's attendance rate, in-class learning situation, and after-class review situation. :

[0133] The formula generated by the deep learning model to quantify a student’s classroom performance is:

[0134] ;

[0135] in, , , is the weight coefficient, Represents the student's classroom performance, Represents the total number of days, Represents different dates in the total number of days, Indicates whether the student is present on different dates. Represents the scores of students' in-class tests on different dates. The score representing the completion of the student's homework;

[0136] The first evaluation report is generated; The numerical value of the first evaluation report is used to generate a quantitative evaluation of the student's classroom performance. The first evaluation report contains the standardized processing value of the score. and evaluation results.

[0137] The second student evaluation model obtains more information about students through the questionnaire set in the data collection module to evaluate students more comprehensively, including;

[0138] Formulate questionnaire questions and evaluation objectives according to the survey content and purpose;

[0139] Pre-process the collected data to ensure its accuracy and completeness;

[0140] Analyze the preprocessed data and extract information related to the evaluation indicators;

[0141] The model is constructed based on the statistical analysis results. The second student evaluation model is constructed using a neural network algorithm and appropriate features are selected for modeling;

[0142] Evaluate and optimize the second student evaluation. The evaluation indicators include precision, recall, and F1 score. Optimize and adjust the second student evaluation model based on the evaluation results.

[0143] Precision ;

[0144] Recall

[0145] ;

[0146] TP stands for True Positive, which is the number of samples correctly predicted as abnormal; FN stands for False Negative, which is the number of samples incorrectly predicted as normal.

[0147] F1 score

[0148] ;

[0149] Where Precision represents the precision rate, and Recall represents the recall rate;

[0150] Generate a second evaluation report, using text generation and visualization to display the relevant data and charts of the second evaluation report. The evaluation report contains the standardized processing value of the score and evaluation results.

[0151] Evaluation integration module: The present invention evaluates students from multiple aspects, and it is necessary to integrate multiple evaluation results to obtain a more accurate, scientific and objective comprehensive evaluation report of students. Therefore, the entropy weight method is used to integrate the learning situation analysis model, the first student evaluation model, and the second student evaluation model, including:

[0152] The evaluation results output by the learning situation analysis model , the standardized processing value of the first evaluation report output by the first student evaluation model , the standardized value of the score in the second evaluation report output by the second student evaluation model Integrated into , by scaling the range, we can get the relative index value of the i-th index ;

[0153] Calculate information entropy For each indicator i, calculate its information entropy ,

[0154] ;

[0155] Calculate weight: According to the calculation results of information entropy, get the information entropy weight of each indicator ,

[0156] ;

[0157] Normalize the calculated weights to obtain the final indicator weights;

[0158] The final score W

[0159] ;

[0160] Used to indicate the student's final score.

[0161] The evaluation reports of the learning situation analysis model, the first student evaluation model, the second student evaluation model and the evaluation integration module are displayed visually, so that students, teachers and parents can see the comprehensive evaluation of students more intuitively and carry out targeted teaching. The page can be generated into a PDF file if the user needs it.

[0162] Privacy data security protection: Encrypt sensitive data in the evaluation system, including student scores, performance evaluations, etc., so that only authorized people can view them. Set up a strict access control mechanism to ensure that only authorized teachers, administrators, and students can access specific data, and record data access logs to track data access. Establish a clear privacy protection agreement, clearly define the scope, purpose, and protection measures of data use, and severely punish violations of the privacy agreement. The above measures protect students' privacy to ensure that students' private information will not be leaked and illegally used or sold.

[0163] The above embodiments only express the preferred implementation modes of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present invention. It should be pointed out that, for a person of ordinary skill in the art, several modifications, improvements and substitutions can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention patent shall be subject to the attached claims.

Claims

1. An AI-based teaching quality evaluation system, comprising: Data collection module: data collection, collecting student data; Data preprocessing module: data preprocessing, data cleaning, format conversion, and normalization; Learning situation analysis model: The learning situation analysis model is established, the pre-processed data is input into the model, the future performance is predicted and analyzed, and the evaluation results are output; First student evaluation model: integrate the student attendance date, the student's in-class test score and the student's homework completion status and input them into the first student evaluation model to output the first student evaluation report; Second student evaluation model: collect student information by questionnaire, and use it as data to input into the second student evaluation model, and output the second student evaluation report; Evaluation integration module: Use the entropy weight method to determine the weights of the learning situation analysis model, the first student evaluation model and the second student evaluation model, and integrate the evaluation results of the learning situation analysis model, the first student evaluation model and the second student evaluation model with the evaluation report according to the weights to form an overall evaluation of the students; Visual interface: Use the Python Flask framework to build a visual interface, connect the learning situation analysis model, the first student evaluation model and the second student evaluation model, summarize the generated evaluation report and display relevant data and charts, and save the evaluation page if necessary to generate a PDF file.

2. The AI-based teaching quality evaluation system according to claim 1, characterized in that: The data collection module obtains student evaluations and information from multiple aspects, including: S201 obtains students’ previous test scores and rankings through the school management system; S202 Students fill in the questionnaire set by the evaluation system; S203 queries the student's daily performance record filled in by the teacher.

3. The AI-based teaching quality evaluation system according to claim 2, characterized in that: The data preprocessing module cleans, converts the format, and normalizes the collected data, including: S301 converts all the collected test scores of students into percentage system for the convenience of subsequent analysis; S302 converts the data in the questionnaire into a csv format file; S303 extracts key information from the student's attendance and the student's daily performance records filled in by the teacher to obtain the data required by the model; S304 data cleaning is to delete the data irrelevant to the prediction in the student daily performance records filled in by the teacher through regular matching.

4. The AI-based teaching quality evaluation system according to claim 3, characterized in that: The learning situation analysis model predicts and analyzes future test scores and trends based on past test scores, including: S401 extracts features, extracts relevant features based on previous test scores, including the lowest score, highest score, average score, and score fluctuations. The score fluctuation is determined by calculating the standard deviation and coefficient of variation: in represents the standard deviation, Indicates the scores of all previous exams. represents the average score, Indicates the number of previous exams. Where CV is the coefficient of variation, is the standard deviation, Indicates the number of previous exams; S402: training the learning situation analysis model, using a multivariate linear regression model, and training the learning situation analysis model according to the extracted features; S403 Evaluation and prediction of the learning situation analysis model: Evaluate the model through training results and test results using the test set, adjust the learning situation analysis model parameters and optimize the learning situation analysis model structure, and predict future test scores and output evaluation results. The evaluation results are presented in text and charts, and the evaluation results include: Standardized processing values ​​of scores , predicted scores and score analysis.

5. The AI-based teaching quality evaluation system according to claim 4, characterized in that: The multivariate linear regression model in S402 includes: S4021 Establish a multiple regression model: in is the explained variable, For all the test scores, is the error term, By using the least squares method The estimated value of the student's future performance is then predicted and analyzed. The specific process includes: First, collect all the test scores of previous students as N groups of observations to obtain N groups of independent variable and dependent variable data; Substitute the expression of the multivariate regression model and treat the error term ε as obeying a normal distribution with a mean of 0; Fit the observed values ​​and calculate the model prediction value for each observed value ; Calculate observations With the model prediction value The square difference between them is summed up to get the error sum of squares SSR (Sum of Squares Residuals) By minimizing the sum of squared errors SSR, the estimated values ​​of the regression coefficients are obtained; Using the matrix form of the least squares method, we can obtain the estimated values ​​of the regression coefficients by solving the normal equations. ; Through training on a large number of data sets The specific value of , and then predict and analyze the student's future scores; S4022 Model evaluation: Evaluate the multivariate regression model through MSE, RMSE, and R-squared values ​​to further adjust the parameters of the multivariate regression model and optimize the structure of the multivariate regression model. Where m is the number of samples, is the actual value, is the predicted value, is the average of the actual values.

6. The AI-based teaching quality evaluation system according to claim 5, characterized in that: The first student evaluation model collects the student's attendance date, the student's in-class test score, and the student's homework completion status in the student's daily performance record filled out by the teacher, and uses a deep learning algorithm to conduct an overall evaluation of the student's daily performance in class, including: S601 Data quantification: quantify and classify the student daily performance records filled in by the teacher after the cleaning in S304 to obtain the student's attendance date, the student's in-class test score and the student's homework completion status, wherein the student's attendance date and the student's in-class test score are expressed in numbers, and the student's homework completion status is divided into four levels and expressed in four special scores; S602: Constructing curves: generating the student's attendance curve, the student's in-class test score change curve and the student's homework completion curve according to the time sequence; S603 generates a functional relationship: establishes a corresponding functional relationship between the student's attendance curve, the student's in-class test score change curve, and the student's homework completion curve through a deep learning model and obtains a correlation coefficient, thereby establishing a correlation function between the student's attendance curve, the student's in-class test score change curve, and the student's homework completion curve, and determines the relationship between the student's attendance date, the student's in-class test score, and the student's homework completion status through the correlation function; S604. Evaluation report generation: for students’ classroom performance, text generation and visualization are used to display relevant data and charts of the evaluation results. The first evaluation report contains the standardized processing value of the score. and evaluation results.

7. The AI-based teaching quality evaluation system according to claim 6, characterized in that: The purpose of using the deep learning model to establish the corresponding functional relationship and obtain the correlation coefficient is to confirm the student's attendance rate, in-class learning situation and after-class review situation by clarifying the relationship between the student's attendance date, the student's in-class test score and the student's homework completion situation, and determine the impact of the student's attendance rate, in-class learning situation and after-class review situation on each other, thereby quantifying the functional relationship between the student's classroom performance and the student's attendance rate, in-class learning situation and after-class review situation.

8. The AI-based teaching quality evaluation system according to claim 7, characterized in that: The formula generated by the deep learning model to quantify students’ classroom performance is: in, , , is the weight coefficient, Represents the student's classroom performance, Represents the total number of days, Represents different dates in the total number of days, Indicates whether the student is present on different dates. Represents the scores of students' in-class tests on different dates. The score representing the completion of the student's assignment.

9. The AI-based teaching quality evaluation system according to claim 8, characterized in that: The second student evaluation model obtains student information by setting up a questionnaire and makes an objective evaluation of the students based on the questionnaire filling results, including: S901 Formulate questionnaire questions and evaluation objectives based on the survey content and purpose; S902 pre-processes the collected data to ensure the accuracy and completeness of the data; S903 analyzes the preprocessed data and extracts information related to the evaluation index; S904 constructs a model based on the statistical analysis results. The second student evaluation model is constructed using a neural network algorithm and appropriate features are selected for modeling; S905 evaluates and optimizes the second student evaluation model, where the evaluation indicators include precision, recall, and F1 score, and optimizes and adjusts the second student evaluation model according to the evaluation results; Recall TP stands for True Positive, which is the number of samples correctly predicted as abnormal; FN stands for False Negative, which is the number of samples incorrectly predicted as normal. F1 score in represents abnormal samples, Represents the overall sample, Precision represents the precision, and Recall represents the recall rate; S906 generates an evaluation report, using text generation and visualization to display relevant data and charts of the evaluation results. The evaluation report includes the standardized processing value of the score and evaluation results.

10. The AI-based teaching quality evaluation system according to claim 9, characterized in that: The S6 evaluation integration module uses the entropy weight method to standardize the learning situation analysis model. , the standardized value of the first student evaluation model , the standardized value of the result of the second student evaluation model The integration results in a more accurate and objective evaluation report for students, and the scores of the three models are standardized. By scaling the range, the relative index value of the i-th index is obtained. , the information entropy corresponding to each indicator i for: Final score calculation : in, To calculate the final score, is information entropy; Used to indicate the student's final score.

Images