Modularized examination score analysis and improvement method based on big data

Abstract

The invention discloses a modular examination score analysis and improvement method based on big data, and the method is characterized in that the method comprises the following steps: 1, carrying out the modular classification of examination questions according to the standards of question types, knowledge points, difficulty and the like, and building a corresponding database; 2, collecting error question data of the students, and inputting the error question data into a database; the invention relates to the technical field of modular score analysis and improvement methods, and relates to a modular examination score analysis and improvement method based on big data, so that students can obtain personalized learning suggestions, learning efficiency and scores can be improved, and the students can have better learning experience. Teachers and educational institutions can make more targeted teaching decisions by using the analysis results, the teaching quality is improved, and the accuracy and correlation of the analysis results can be improved along with time by adjusting the regression coefficient of the model and updating the model in real time.

Description

Technical Field

[0001] The present invention relates to the technical field of modular score analysis and improvement methods, and in particular to a modular test score analysis and improvement method based on big data. Background Art

[0002] In existing education assessment systems, test score analysis often lacks pertinence and personalization, and cannot effectively help students identify and improve their weak links in learning. In addition, traditional test analysis methods may not be able to fully utilize big data technology to provide in-depth insights and predictions. Therefore, the present invention aims to provide a more accurate and personalized test score analysis method through modular classification and big data analysis to help students and educators better understand learning outcomes and improve learning efficiency. Summary of the invention

[0003] The present invention provides a modular test score analysis and improvement method based on big data. Students can obtain personalized learning suggestions, which helps to improve their learning efficiency and grades. Teachers and educational institutions can use these analysis results to make more targeted teaching decisions and improve teaching quality. By adjusting the regression coefficient of the model and updating the model in real time, it can ensure that the accuracy and relevance of the analysis results improve over time.

[0004] A modular test score analysis and improvement method based on big data, characterized in that the method comprises the following steps:

[0005] Step 1: Classify the test questions into modular categories according to question types, knowledge points, difficulty and other standards, and establish a corresponding database;

[0006] Step 2: Collect students’ wrong answer data and input them into the database;

[0007] Step 3: Use algorithm S1 to analyze wrong questions, compare and analyze the wrong questions of students with the modular questions in the database, use the correlation analysis algorithm to calculate the correlation coefficient between the wrong questions of students and different modules, determine the module to which the wrong questions belong, and screen out the common wrong questions of other students with similar scores to the student;

[0008] Step 4: Use algorithm S2 to perform multivariate correlation analysis, select modules in the set of wrong questions from students with similar total scores, and sort them;

[0009] Step 5: Use algorithm S3 to dynamically analyze students’ wrong questions according to time series, calculate the fixed-base ratio and year-on-year ratio, and analyze the dynamic change trend of wrong question scores;

[0010] Step 6: Use algorithm S4 to perform regression analysis to predict the student's error rate for each module in the next exam and generate a predicted score;

[0011] As a further limitation of the technical solution, the modular classification further includes: modular classification of all test questions according to big data; according to the mathematical field to which they belong, they are divided into algebra questions, plane geometry questions, and trigonometry questions; according to the elements of the questions, they are divided into standard questions, training questions, exploration questions, and problem questions; according to the question type, the test questions are divided into fill-in-the-blank questions, judgment questions, multiple-choice questions, calculation questions, application questions, proof questions, drawing questions, thinking questions, reading questions, regularity questions, and answer questions; according to the attribution of the main knowledge points, they are divided into the first unit, the second unit, the third unit...the Nth unit;

[0012] The error analysis includes: modularizing each test paper of the students into a database; classifying all the error questions of the test papers of the subject in the entire learning stage of different students into the entire database;

[0013] The multivariate correlation analysis includes: calculating the matrix formed by the correlation coefficients between any two variables in different wrong questions b, c, d, e; sorting the different combinations according to the size of R;

[0014] The dynamic analysis includes: finding the corresponding index values ​​of the factors in each period; finding the corresponding ratios between them; analyzing the dynamic change trend of the index factors through the changes in the found ratios;

[0015] The regression analysis includes: using the regression model to predict the error rate of each module of the student in the next exam; using the F test to test the significance of the model; adjusting the regression coefficient of the model in real time and updating the coefficient of the model in real time.

[0016] As a further limitation of the present technical solution, the algorithm S1 is a correlation analysis algorithm, and the formula for performing wrong question analysis using the algorithm S1 is as follows:

[0017]

[0018] Among them, r xs is sample data 1, and r ys For sample data 2, is the mean of sample data 1, is the mean of sample data 2, sim(x,y) is the correlation coefficient r, Sxy is the total data set, and s refers to one of the data, which belongs to the data set Sxy.

[0019] The correlation coefficient r is a unitless variable with a value range of -1≤r≤1. The closer the test result r is to 1, the closer the linear correlation between the variables is; conversely, the closer |r| is to 0, the more distant the linear relationship is. If the r value is positive, it indicates that the two variables are positively correlated; if the r value is negative, it indicates that the two variables are negatively correlated.

[0020] When analyzing wrong questions, the data of wrong questions a, b, c, d, e, etc. are analyzed for correlation with the data of different modules in the database, and the correlation coefficients of different wrong questions in different modules are calculated respectively, so that the module with the largest correlation coefficient is assigned to complete the classification and classification of different wrong questions;

[0021] Calculate the correlation coefficient r between student a's score and the scores of different students. The student with the largest r is considered to be the closest and most relevant to student a. At the same time, count all their wrong questions, organize the different range modules to which the wrong questions belong, and form the result.

[0022] As a further limitation of the present technical solution, the algorithm S2 is a multivariate correlation analysis algorithm. The types of wrong questions of student a are analyzed and sorted according to the multivariate correlation analysis algorithm. The matrix formed by the correlation coefficients between any two variables in different wrong questions b, c, d, and e is:

[0023]

[0024] where r ij is the simple correlation coefficient between any two variables:

[0025]

[0026] Among them, R is the correlation coefficient, p is one of the variables, and x i For sample data 1, is the mean of sample data 1, y i For sample data 2, is the mean of sample data 2, i is the type of different sample data, j is the type of different sample data, r pn is the correlation coefficient between sample p and sample n, r np is the correlation coefficient between sample n and sample p.

[0027] The correlation coefficient R of different wrong question classification modules is calculated according to the above formula, and the different combination classification arrangements are sorted according to the size of R, and each different classification arrangement is listed.

[0028] As a further limitation of the present technical solution, the algorithm S3 is a dynamic analysis method, which is also called a time series analysis method. First, the corresponding index values ​​of the factors in each period are obtained, and the corresponding proportions between them are obtained. Then, the dynamic change trend of the index factors is analyzed by the obtained proportion changes.

[0029] Count the students' wrong questions in chronological order;

[0030] The statistical calculation is performed based on the data of different periods such as monthly exams, midterm exams, and final exams. The calculation formula for the fixed base ratio is:

[0031]

[0032] Where q o is the cardinality, q i Indicates the value of the comparison period (i=1,2,...n).

[0033] The calculation formula for the month-on-month increase is:

[0034]

[0035] Where q i-1 Represents the comparison period q i The value of the previous period (i=1,2,...n).

[0036] According to the above formula, calculate the dynamic changes in students' grades and wrong questions in the entire academic year or semester, make a dynamic series chart of the average, and make a base ratio and year-on-year dynamic chart for further analysis.

[0037] The average dynamic series chart takes time as the horizontal axis variable and the average of scores in different periods as the vertical axis variable to establish the average dynamic series chart.

[0038] For the fixed base ratio and month-on-month dynamic chart, take time as the horizontal axis variable, and take the fixed base ratio value and month-on-month value as the vertical axis variables to make a line chart.

[0039] Through dynamic series graphs, fixed base ratios, and month-on-month dynamic graphs, we analyze the indicator values ​​of students' different wrong questions in different periods, and analyze the dynamic change trend of wrong question scores through proportional changes, so as to focus on the classification question types with the largest fluctuations and the corresponding least mature knowledge points, and conduct focused study and review to achieve targeted improvement.

[0040] As a further limitation of the present technical solution, the algorithm S4 is a regression analysis algorithm,

[0041] The regression model is:

[0042] y=A1+A2*x1+...+A3*x2+A n *x n +e

[0043]

[0044] Where y is the actual result, A n is the correlation coefficient of different data, x1,x x ,...,x n is the sample data, is the prediction result, a1,a2,...,a n is the correlation coefficient of different data, and e is the error adjustment value

[0045] Correlation analysis between y and x1, x2, .... Correlation analysis;

[0046] The correlation coefficient calculation formula is:

[0047]

[0048] According to the regression model and correlation coefficient, the error rate of each module of the student's next exam is predicted, where R is the correlation coefficient, y is the actual calculated error rate, and y i is the sample data, is the error rate predicted by the model, variables x1,x2,...,x n It is the category of wrong questions in different modules, that is, the variable. The model coefficient is calculated through each data, and the F test is used to test the significance of the model, and the regression coefficient of the model is adjusted in real time, and the coefficient of the model is updated in real time to make the model more accurate.

[0049] A computer program product for a modular test score analysis and improvement method based on big data, the program product comprising a computer-readable storage medium and a computer program stored on the medium, wherein the program is designed to execute the steps of the method described in any one of claims 1 to 7.

[0050] A computer system for executing a modular test score analysis and improvement method based on big data, comprising: a storage module and a processing module, wherein the storage module is used to store test questions and students' wrong test data, and the processing module is used to execute the steps of the method described in any one of claims 1 to 6.

[0051] Compared with the prior art, the advantages and positive effects of the present invention are:

[0052] By modularizing the test questions according to question type, knowledge points, difficulty and other criteria, we can analyze students' wrong questions more systematically, thereby more accurately identifying weak links in learning;

[0053] Using big data analysis, we can process and analyze a large amount of student error data to discover broader patterns and trends and provide students with more personalized learning suggestions;

[0054] By calculating the correlation coefficient between wrong questions and different modules, the module to which the wrong questions belong can be determined, which helps students focus on the areas that need the most improvement;

[0055] Through multivariate correlation analysis, the modules in the set of wrong questions can be screened and sorted from students with similar scores, which helps to find which modules are common difficulties for most students;

[0056] Through time series analysis, we can observe the trend of students' performance changes in different time periods, so as to provide students with timely learning feedback and intervention;

[0057] Use regression analysis to predict the student's error rate for each module in the next exam and generate a predicted score, which helps students and teachers understand the possible exam performance in advance and adjust their study plans accordingly;

[0058] Through the above analysis, students can get personalized learning suggestions to help improve their learning efficiency and grades. Teachers and educational institutions can use these analysis results to make more targeted teaching decisions and improve teaching quality. By adjusting the regression coefficients of the model and updating the model in real time, the accuracy and relevance of the analysis results can be ensured to improve over time. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0060] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0061] The present invention will be further described below in conjunction with the accompanying drawings and implementation modes.

[0062] A modular test score analysis and improvement method based on big data, the method comprising the following steps:

[0063] Step 1: modularize the test questions according to question type (such as fill-in-the-blank questions, multiple-choice questions, etc.), knowledge points (such as Unit 1, Unit 2, etc.), difficulty, etc., and establish a corresponding database to store these modular questions;

[0064] Step 2: Collect students’ wrong answer data and input them into the database;

[0065] Among them, after each test, the students' wrong question data are collected and input into the established database;

[0066] Step 3: Use algorithm S1 to analyze wrong questions, compare and analyze the wrong questions of students with the modular questions in the database, use the correlation analysis algorithm to calculate the correlation coefficient between the wrong questions of students and different modules, determine the module to which the wrong questions belong, and filter out the common wrong questions of other students with similar scores to the student;

[0067] Among them, the wrong questions b, c, d, e, etc. of student a are classified into the entire database for comparison, and the range of common wrong questions among students with a total score close to that of student a is selected and pushed to the user terminal, and is pushed to the first user terminal first, and the first user terminal is preferably a teacher terminal;

[0068] Step 4: Use algorithm S2 to perform multivariate correlation analysis, select modules in the set of wrong questions from students with similar total scores, and sort them;

[0069] Among them, the modules in the wrong questions set are selected from students with similar total scores, the matrix composed of the correlation coefficients between any two variables between different wrong questions is calculated, and different combinations are classified and arranged according to the size of the correlation coefficient R;

[0070] Among them, students with a total score close to that of student a are selected, and among various classification methods, the classification types with the most concentration of student a are arranged from top to bottom. For example, among wrong questions b, c, d, and e, b, c, and d are all drawing questions, and they have the most errors in the same category. The algorithm puts this arrangement at the front and releases each arrangement method in turn, so that different students can generate their own unique key learning modules according to targeted classification. The module can be multiple-choice questions, true-or-false questions, etc., or it can be the first unit, the second unit, etc.

[0071] Step 5: Use algorithm S3 to dynamically analyze students' wrong questions according to time series (such as monthly exams, midterm exams, and final exams), calculate the fixed-base ratio and year-on-year ratio, and analyze the dynamic change trend of wrong question scores;

[0072] Among them, all the wrong questions in the test papers of this subject during the student's entire learning stage are summarized, and a curve graph is generated according to each category based on the time axis, and the ones with the largest peak fluctuations are arranged in the front, so as to obtain a curve graph of the rising and falling fluctuations of the wrong ratio of students in each category, such as multiple-choice questions, during this learning stage. The graph with the largest fluctuation is arranged in the front of the user terminal for special attention, indicating that the mastery of this type of questions is unstable and the reasons need to be analyzed and focused on learning and mastering them firmly.

[0073] Summarize all wrong questions in the subject during the entire learning period of the student, generate a curve chart according to each category based on the time axis, and arrange the ones with the largest fluctuation in the peak of the curve at the front, so as to obtain a curve chart of the fluctuation of the wrong ratio of each category, such as multiple-choice questions, in this learning period of the student. The one with the largest fluctuation is arranged at the front, and the focus is on this type of question, which is unstable to master. Analyze the reasons and focus on learning and mastering it firmly to generate a curve chart of the change of grades;

[0074] Step 6: Use algorithm S4 to perform regression analysis to predict the student's error rate for each module in the next exam and generate a predicted score;

[0075] Among them, the regression model is used to predict the error rate of students in each module in the next exam, the F test is used to test the significance of the model, the regression coefficient of the model is adjusted in real time, and the coefficient of the model is updated in real time.

[0076] Among them, the entire test paper of the next exam is input, all test questions are classified and delineated, the average error rate and median error rate are calculated according to the students' overall mastery of this type, and a curve graph is generated in combination with the time axis. The curve peak is calculated by algorithm s4, and each module is comprehensively calculated to obtain the student's predicted score for the next exam, and the error rate of each module predicted for the student's next exam is obtained. Multiple calculations are performed and compared with the actual situation, so that the next calculation will be more accurate and intelligent.

[0077] The modular classification further includes: modular classification of all test questions according to big data; according to the mathematics field to which they belong, they are divided into algebra questions, plane geometry questions, and trigonometry questions; according to the elements of the questions, they are divided into standard questions, training questions, exploration questions, and problem questions; according to the question type, the test questions are divided into fill-in-the-blank questions, judgment questions, multiple-choice questions, calculation questions, application questions, proof questions, drawing questions, thinking questions, reading questions, regularity questions, and answer questions; according to the attribution of the main knowledge points, they are divided into the first unit, the second unit, the third unit...the Nth unit;

[0078] The error analysis includes: modularizing each test paper of the students into a database; classifying all the error questions of the test papers of the subject in the entire learning stage of different students into the entire database;

[0079] The multivariate correlation analysis includes: calculating the matrix formed by the correlation coefficients between any two variables in different wrong questions b, c, d, e; sorting the different combinations according to the size of R;

[0080] The dynamic analysis includes: finding the corresponding index values ​​of the factors in each period; finding the corresponding ratios between them; analyzing the dynamic change trend of the index factors through the changes in the found ratios;

[0081] The regression analysis includes: using the regression model to predict the error rate of each module of the student in the next exam; using the F test to test the significance of the model; adjusting the regression coefficient of the model in real time and updating the coefficient of the model in real time.

[0082] The algorithm S1 is a correlation analysis algorithm;

[0083] Correlation is the use of numbers to show the correlation between changes in two or more variables. It is usually used to express a certain statistical regularity between variables, where one variable changes roughly in accordance with a certain trend as another variable or a group of variables changes. Correlation is not an exact linear relationship between variables, but an empirical relationship established after a large number of observations or experiments.

[0084] Correlation analysis is to study whether there is a certain correlation between the variables, and to explore the correlation direction and degree of correlation for specific phenomena caused by dependence, and to use the correlation coefficient r to show whether there is a close relationship between the variables and the degree of closeness.

[0085] The formula for error analysis in the algorithm S1 is as follows:

[0086]

[0087] Among them, r xs and r ys They are two different data. and is the sample mean of different samples, sim(x,y) is the correlation coefficient r, Sxy is the total data set, and s refers to one of the data, which belongs to the data set Sxy.

[0088] The correlation coefficient r is a unitless variable with a value range of -1≤r≤1. The closer the test result r is to 1, the closer the linear correlation between the variables is; conversely, the closer |r| is to 0, the more distant the linear relationship is. If the r value is positive, it indicates that the two variables are positively correlated; if the r value is negative, it indicates that the two variables are negatively correlated.

[0089] When analyzing wrong questions, the data of wrong questions a, b, c, d, e, etc. are analyzed for correlation with the data of different modules in the database, and the correlation coefficients of different wrong questions in different modules are calculated respectively, so that the module with the largest correlation coefficient is assigned to complete the classification and classification of different wrong questions;

[0090] Calculate the correlation coefficient r between student a's score and the scores of different students. The student with the largest r is considered to be the closest and most relevant to student a. At the same time, count all their wrong questions, organize the different range modules to which the wrong questions belong, and form the result.

[0091] The algorithm S2 is a multivariate correlation analysis algorithm. Multivariate correlation is also called complex correlation, which indicates the total correlation between m independent variables and dependent variables. The multivariate correlation coefficient is also called complex correlation coefficient, which indicates the total closeness between m independent variables and dependent variables. It is represented by R, and its value range is [0,1]. The closer the r value is to 1, the closer the total correlation is, and the closer it is to 0, the less close the total correlation is. The hypothesis test of R is generally tested by F test. The square of multivariate correlation coefficient R 2 It is called the coefficient of determination and represents the ratio of the sum of squares of the multiple regression to the sum of squares of the total variation in y.

[0092] According to the multivariate correlation analysis algorithm, the types of wrong questions of student a are analyzed and sorted out. The matrix composed of the correlation coefficients between any two variables in different wrong questions b, c, d, and e is:

[0093]

[0094] where r i j is the simple correlation coefficient between any two variables:

[0095]

[0096] R is the correlation coefficient, p is one of the variables, xi is sample data 1, is the mean of sample data 1, y i For sample data 2, is the mean of sample data 2, i is the type of different sample data, j is the type of different sample data, r pn is the correlation coefficient between sample p and sample n, r np is the correlation coefficient between sample n and sample p.

[0097] The correlation coefficient R of different wrong question classification modules is calculated according to the above formula, and the different combination classification arrangements are sorted according to the size of R, and each different classification arrangement is listed.

[0098] The algorithm S3 is a dynamic analysis method, which is also called a time series analysis method. First, the corresponding index values ​​of the factors in each period are obtained, and the corresponding proportions between them are obtained. Then, the dynamic change trend of the index factors is analyzed through the obtained proportion changes.

[0099] Count the students' wrong questions in chronological order;

[0100] The statistical calculation is performed based on the data of different periods such as monthly exams, midterm exams, and final exams. The calculation formula for the fixed base ratio is:

[0101]

[0102] Where q o is the cardinality, q i Indicates the value of the comparison period (i=1,2,...n).

[0103] The calculation formula for the month-on-month increase is:

[0104]

[0105] Where q i-1 Represents the comparison period q i The value of the previous period (i=1,2,...n).

[0106] According to the above formula, calculate the dynamic changes in students' grades and wrong questions in the entire academic year or semester, make a dynamic series chart of the average, and make a base ratio and year-on-year dynamic chart for further analysis.

[0107] The average dynamic series chart takes time as the horizontal axis variable and the average of scores in different periods as the vertical axis variable to establish the average dynamic series chart.

[0108] For the fixed base ratio and month-on-month dynamic chart, take time as the horizontal axis variable, and take the fixed base ratio value and month-on-month value as the vertical axis variables to make a line chart.

[0109] Through dynamic series graphs, fixed base ratios, and month-on-month dynamic graphs, we analyze the indicator values ​​of students' different wrong questions in different periods, and analyze the dynamic change trend of wrong question scores through proportional changes, so as to focus on the classification question types with the largest fluctuations and the corresponding least mature knowledge points, and conduct focused study and review to achieve targeted improvement.

[0110] The algorithm S4 is a regression analysis algorithm. Regression analysis is an analysis method that studies whether there is a causal relationship between two or more variables. On the basis of correlation analysis, the general relationship between the quantitative changes of two or more variables with a correlation is measured to determine a suitable mathematical model in order to infer another unknown quantity from one known quantity.

[0111] The regression model is:

[0112] y=A1+A2*x1+...+A3*x2+A n *x n +e

[0113]

[0114] Where y is the actual result, A n is the correlation coefficient of different data, x1,x x ,...,x n is the sample data, is the prediction result, a1,a2,...,a n is the correlation coefficient of different data, and e is the error adjustment value

[0115] Correlation analysis between y and x1, x2, .... Correlation analysis;

[0116] The correlation coefficient calculation formula is:

[0117]

[0118] According to the regression model and correlation coefficient, the error rate of each module of the student's next exam is predicted, where R is the correlation coefficient, y is the actual calculated error rate, and y i is the sample data, is the error rate predicted by the model, variables x1,x2,...,x n It is the category of wrong questions in different modules, that is, the variable. The model coefficient is calculated through each data, and the F test is used to test the significance of the model, and the regression coefficient of the model is adjusted in real time, and the coefficient of the model is updated in real time to make the model more accurate.

[0119] The present invention also provides a computer program product for a modular test score analysis and improvement method based on big data, the program product comprising a computer-readable storage medium and a computer program stored on the medium, the program being designed to execute the steps of the method described in any one of claims 1 to 7.

[0120] The present invention also provides a computer system for executing a modular test score analysis and improvement method based on big data, comprising: a storage module for storing test questions and students' wrong question data;

[0121] A processing module, used to execute the steps of the method described above;

[0122] The processing module also includes a first user terminal, a second user terminal and a network server. The first user terminal is preferably a teacher terminal, and the second user terminal is preferably a student terminal. The user terminals include one or more of smart phones, personal computers or handheld computers with communication functions.

[0123] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A modular test score analysis and improvement method based on big data, characterized by: The method comprises the following steps: Step 1: Classify the test questions into modular categories according to question type, knowledge points, and difficulty, and establish a corresponding database; Step 2: Collect students’ wrong answer data and input them into the database; Step 3: Use algorithm S1 to analyze wrong questions, compare and analyze the wrong questions of students with the modular questions in the database, use the correlation analysis algorithm to calculate the correlation coefficient between the wrong questions of students and different modules, determine the module to which the wrong questions belong, and filter out the common wrong questions of other students with similar scores to the student; Step 4: Use algorithm S2 to perform multivariate correlation analysis, select modules in the set of wrong questions from students with similar total scores, and sort them; Step 5: Use algorithm S3 to dynamically analyze students’ wrong questions according to time series, calculate the fixed-base ratio and year-on-year ratio, and analyze the dynamic change trend of wrong question scores; Step 6: Use algorithm S4 to perform regression analysis to predict the student's error rate for each module in the next exam and generate a predicted score.

2. The modular test score analysis and improvement method based on big data according to claim 1 is characterized by: The modular classification further includes: modular classification of all test questions according to big data; according to the mathematics field to which they belong, they are divided into algebra questions, plane geometry questions, and trigonometry questions; according to the elements of the questions, they are divided into standard questions, training questions, exploration questions, and problem questions; according to the question type, the test questions are divided into fill-in-the-blank questions, judgment questions, multiple-choice questions, calculation questions, application questions, proof questions, drawing questions, thinking questions, reading questions, regularity questions, and answer questions; according to the attribution of the main knowledge points, they are divided into the first unit, the second unit, the third unit...the Nth unit; The error analysis includes: modularizing each test paper of the students into a database; classifying all the error questions of the test papers of the subject in the entire learning stage of different students into the entire database; The multivariate correlation analysis includes: calculating the matrix formed by the correlation coefficients between any two variables in different wrong questions b, c, d, e; sorting the different combinations according to the size of R; The dynamic analysis includes: finding the corresponding index values ​​of the factors in each period; finding the corresponding ratios between them; analyzing the dynamic change trend of the index factors through the changes in the found ratios; The regression analysis includes: using the regression model to predict the error rate of each module of the student in the next exam; using the F test to test the significance of the model; adjusting the regression coefficient of the model in real time and updating the coefficient of the model in real time.

3. The modular test score analysis and improvement method based on big data according to claim 2 is characterized by: The algorithm S1 is a correlation analysis algorithm. The formula for analyzing wrong questions using the algorithm S1 is as follows: Among them, r xs and r ys They are two different data. and is the sample mean of different samples, sim(x,y) is the correlation coefficient r, Sxy is the total data set, and s refers to one of the data, which belongs to the data set Sxy. The correlation coefficient r is a unitless variable with a value range of -1≤r≤1. The closer the test result r is to 1, the closer the linear correlation between the variables is; conversely, the closer |r| is to 0, the more distant the linear relationship is. If the r value is positive, it indicates that the two variables are positively correlated; if the r value is negative, it indicates that the two variables are negatively correlated. When analyzing wrong questions, the data of wrong questions a, b, c, d, e, etc. are analyzed for correlation with the data of different modules in the database, and the correlation coefficients of different wrong questions in different modules are calculated respectively, so that the module with the largest correlation coefficient is assigned to complete the classification and classification of different wrong questions; Calculate the correlation coefficient r between student a's score and the scores of different students. The student with the largest r is considered to be the closest and most relevant to student a. At the same time, count all their wrong questions, organize the different range modules to which the wrong questions belong, and form the result.

4. The modular test score analysis and improvement method based on big data according to claim 3 is characterized by: The algorithm S2 is a multivariate correlation analysis algorithm. The types of wrong questions of student a are analyzed and sorted according to the multivariate correlation analysis algorithm. The matrix formed by the correlation coefficients between any two variables in different wrong questions b, c, d, and e is: where r ij is the simple correlation coefficient between any two variables: Among them, R is the correlation coefficient, p is one of the variables, and x i To represent sample data 1, is the mean of sample data 1, y i To represent sample data 2, is sample data 2, i is the type of different sample data, j is the type of different sample data, r pn is the correlation coefficient between sample p and sample n, r np is the correlation coefficient between sample n and sample p. The correlation coefficient R of different wrong question classification modules is calculated according to the above formula, and the different combination classification arrangements are sorted according to the size of R, and each different classification arrangement is listed.

5. The modular test score analysis and improvement method based on big data according to claim 4 is characterized by: The algorithm S3 is a dynamic analysis method, which is also called a time series analysis method. First, the corresponding index values ​​of the factors in each period are obtained, and the corresponding proportions between them are obtained. Then, the dynamic change trend of the index factors is analyzed through the obtained proportion changes. Count the students' wrong questions in chronological order; The statistical calculation is performed based on the data of different periods such as monthly exams, midterm exams, and final exams. The calculation formula for the fixed base ratio is: Where q o is the cardinality, q i Indicates the value of the comparison period (i=1,2,...n). The calculation formula for the month-on-month increase is: Where q i-1 Represents the comparison period q i The value of the previous period (i=1,2,...n). According to the above formula, the dynamic changes in students' grades and wrong questions in a whole school year or semester are calculated, and further analysis is made. The indicator values ​​of students' wrong questions in different periods are analyzed, and the dynamic change trend of wrong question scores is analyzed through proportional changes, so as to focus on the classification questions with the largest fluctuations and the corresponding least mature knowledge points, and carry out focused study and review to achieve targeted improvement.

6. The modular test score analysis and improvement method based on big data according to claim 5 is characterized by: The algorithm S4 is a regression analysis algorithm. The regression model is: <h2 style=";text-align:left;direction:ltr">y = A1 + A2 * x1 + ... + A3 * x2 + A<h2 style=";text-align:left;direction:ltr"> n <h2 style=";text-align:left;direction:ltr"> *x<h2 style=";text-align:left;direction:ltr"> n <h2 style=";text-align:left;direction:ltr"> +e Where y is the actual result, A n is the correlation coefficient of different data, x1,x x ,...,x n is the sample data, is the prediction result, a1,a2,...,a n is the correlation coefficient of different data, and e is the error adjustment value Correlation analysis between y and x1, x2, .... Correlation analysis; The correlation coefficient calculation formula is: According to the regression model and correlation coefficient, the error rate of each module of the student's next exam is predicted, where R is the correlation coefficient, y is the actual calculated error rate, and y i is the sample data, is the error rate predicted by the model, variables x1,x2,...,x n It is the category of wrong questions in different modules, that is, the variable. The model coefficient is calculated through each data, and the F test is used to test the significance of the model, and the regression coefficient of the model is adjusted in real time, and the coefficient of the model is updated in real time to make the model more accurate.

7. A computer program product for a modular test score analysis and improvement method based on big data according to any one of claims 1 to 6, characterized in that: The program product comprises a computer-readable storage medium and a computer program stored on the medium, wherein the program is designed to execute the steps of the method according to any one of claims 1 to 6.

8. A computer system for executing a modular test score analysis and improvement method based on big data according to any one of claims 1 to 6, characterized in that: It comprises a storage module and a processing module, wherein the storage module is used to store test questions and students' wrong answer data, and the processing module is used to execute the steps of the method described in any one of claims 1 to 6.

Images