Education quality evaluation method based on machine learning model

Abstract

The invention discloses an education quality evaluation method based on a machine learning model, and particularly relates to the technical field of education quality evaluation. The change condition of the characteristic parameters of each region at different time points is judged through an analytic hierarchy process, the weight assignment of each region is determined, the overall educational data quality index is calculated, and the change trend of the educational data is analyzed through a time sequence analysis method; after the overall education data quality index and the prediction performance change condition of the model during new data processing are analyzed by using fuzzy logic, the prediction capability of the model in different regions and education environments is evaluated, the model is correspondingly improved, the accuracy of education data quality evaluation and prediction model is effectively improved, and the prediction efficiency is improved. Through careful classification and improvement measures, accurate optimization of model prediction capability in different areas and education environments is realized, and reasonable configuration of education resources and continuous improvement of teaching quality are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of education quality assessment, and in particular to an education quality assessment method based on a machine learning model. Background Art

[0002] Education quality assessment based on machine learning models refers to the use of machine learning algorithms and data analysis techniques to conduct in-depth analysis of various types of data in the education system (such as student grades, teacher teaching effectiveness, learning resource usage, etc.) to evaluate and improve education quality. Through training on a large amount of historical data, machine learning models can discover hidden patterns and relationships and provide quantitative evaluation and prediction of education quality. This evaluation method can more accurately and comprehensively reflect the actual situation of education quality, reduce human bias, and can be updated and optimized in real time. Through automated data processing and analysis, educational institutions can identify problems in a timely manner, formulate effective improvement measures, improve the overall education level, and ultimately achieve the goals of personalized teaching and equitable education.

[0003] The prior art has the following deficiencies:

[0004] The generalization ability of machine learning models may vary in different regions and educational environments. If the data used in model training mainly comes from a specific region or educational background, the evaluation results may not have good predictive ability for the quality of education in other regions. In particular, for some regions with relatively scarce educational resources or different educational models, the evaluation results of the model may be seriously distorted. This will lead to poor implementation of educational policies and even further aggravate educational disparities between regions. Summary of the invention

[0005] The purpose of the present invention is to provide an education quality assessment method based on a machine learning model to address the deficiencies in the background technology.

[0006] In order to achieve the above object, the present invention provides the following technical solution: an education quality assessment method based on a machine learning model, comprising the following steps:

[0007] S1: Determine the criteria for different regions and educational settings, and collect educational data from several regions and educational settings at multiple time points;

[0008] S2: Preprocess the education data from different sources and extract the characteristic parameters of the education data at multiple time points in different regions based on the evaluation criteria of education quality;

[0009] S3: Use the analytic hierarchy process to determine the changes in the characteristic parameters of each region at different time points, determine the weight assignment of each region, and calculate the overall education data quality index after weighted average calculation of the weight assignment of each region;

[0010] S4: Analyze the changing trends of educational data through time series analysis and evaluate the changes in the model's predictive performance when processing new data;

[0011] S5: After analyzing the overall education data quality index and the changes in the model's predictive performance when processing new data through fuzzy logic, the model's predictive ability in different regions and educational settings was evaluated;

[0012] S6: Based on the evaluation results, the predictive ability of the model in different regions and educational environments is divided into high predictive ability, medium predictive ability and low predictive ability, and the model is improved accordingly based on the division results.

[0013] In a preferred embodiment, in S2, characteristic parameters in the education data of each region and time point are obtained, including the incremental index of teaching resource usage and the abnormal index of teacher turnover rate.

[0014] In a preferred embodiment, the method for obtaining the incremental index of teaching resource usage is:

[0015] Obtain the teaching resource usage data at the same time point t in each region, collect the independent variable data that affect the use of teaching resources, and ensure the integrity and consistency of the data;

[0016] According to the obtained teaching resource usage data, a multivariate linear regression model is selected, and the regression model is trained using historical data to fit the model parameters. The specific calculation expression of the multivariate linear regression model is: T t =(s0+s1X 1t +s2X 2t +...+s n X nt )+c; where T t represents the usage of teaching resources at time point t, X 1t , X 2t , ..., X nt are the influencing factors of the independent variable at time point t, s0, s1, s2, ..., s n is the regression coefficient, c is the error term, and the predicted teaching resource usage T at each time point is calculated. t ,

[0017] The teaching resource usage in the time period t+1 and the teaching resource usage in the time period t are analyzed, and the teaching resource usage increment index is calculated. The teaching resource usage increment index is obtained by calculating the ratio of the difference between the teaching resource usage in the time period t+1 and the teaching resource usage in the time period t and marking it as LB.

[0018] In a preferred embodiment, the method for obtaining the abnormal index of teacher turnover rate is:

[0019] Collect teacher turnover rate data at various time points and regions, clean and standardize the data, and construct the prior distribution P(μ,σ 2 ), μ is the posterior mean, which represents the expected value of the teacher turnover rate, and σ is the posterior standard deviation, which represents the standard deviation of the teacher turnover rate;

[0020] According to historical data, the likelihood function of the turnover rate is established to calculate the probability distribution of the turnover rate under given data conditions. The specific calculation expression is: In the formula, P(R|μ,σ 2 ) is the probability distribution of the turnover rate, R is the actual observed teacher turnover rate, and the posterior distribution is calculated using the Bayesian theorem. The calculation expression is: P(μ,σ 2 ∣R)∝P(R∣μ,σ 2 )P(μ,σ 2 ); P(μ,σ 2 |R) is the probability of abnormal situation under the condition of turnover rate R. The abnormal situation of teacher turnover rate is evaluated according to the posterior distribution. By calculating the posterior probability, the abnormal points are identified and measured, and the abnormal index of teacher turnover rate is calculated. The specific calculation expression is: Where BC is the abnormal index of teacher turnover rate.

[0021] In a preferred embodiment, in S3, a hierarchical model is constructed to represent the goals, criteria and plans in a hierarchical manner, which is divided into three levels: the goal layer, the criterion layer and the plan layer; the goal layer is the overall education data quality index, the criterion layer is the incremental index of teaching resource use and the abnormal index of teacher turnover rate, and the plan layer is the data of different regions and time points. Judgment matrices are constructed for the criterion layer and the plan layer respectively for comparison to determine the importance between the incremental index of teaching resource use and the abnormal index of teacher turnover rate, the criterion layer matrix A is normalized, and the eigenvector W of the matrix is ​​calculated. A ;

[0022] For each region, the judgment matrix of the scheme layer is constructed based on the incremental index of teaching resource use and the abnormal index of teacher turnover rate, which are represented as B1 and B2 respectively. Each judgment matrix B1 and B is normalized to calculate the eigenvector W B1 and W B2 ;

[0023] Combine the weight vectors of the criterion layer and the solution layer to calculate the comprehensive weight of each region, W 总 =W A *(W B1 , W B2 );W总 As the comprehensive weight of each region, the weight values ​​of each region are assigned and weighted averaged to obtain the overall education data quality index.

[0024] In a preferred embodiment, in S4, the changing trend of the education data is analyzed by time series analysis method, specifically:

[0025] Collect historical education data, including student grades, teacher evaluations, school facilities, and teaching resources, and build a time series with the collected data. Decompose the time series data into trend, seasonal, and random components, and select the moving average time series model ARIMA. The calculation expression of the model is: ec k =h+(v1G k-1 +v2G k-2 +...+v p G k-p +m1ec k-1 +m2ec k-2 +...+m p ec k-p );where ec k is the observed value at time point k, h is a constant term, which represents the average level of the sequence, v1, v2, ..., v p are the autoregressive coefficients, m1, m2, ..., m p is the moving average coefficient, ec k-1 ,ec k-2 ,...,ec k-p is the historical data value, G k-1 , G k-2 , ..., G k-p is the error term;

[0026] Use historical data to train the selected time series model, fit the model parameters, split the data set into training set and test set, verify the model's predictive performance by predicting the test set data, use the trained model to predict future education data, compare and analyze the actual data and predicted data, calculate the standard deviation between the actual data and the predicted data, record it as the education data deviation value, and use the education data deviation value to evaluate the change in the model's predictive performance when processing new data.

[0027] In a preferred embodiment, the acquired education data deviation value is compared with a reference threshold of the education data deviation value under a standard state set in the historical data. If the education data deviation value is greater than or equal to the reference threshold of the education data deviation value under a standard state set in the historical data, a model prediction performance abnormality signal is generated; if the education data deviation value is less than the reference threshold of the education data deviation value under a standard state set in the historical data, a model prediction performance normal signal is generated.

[0028] In a preferred embodiment, in S5, the overall education data quality index and the education data deviation value are used as input items, and the prediction ability of the model is used as the output item;

[0029] Define fuzzy sets for each input and output variable and choose membership functions;

[0030] Construct fuzzy rules to map input variables to output variables;

[0031] Using Mamdan i reasoning, the membership of input variables is applied to fuzzy rules, the fuzzy output is calculated for the given overall education data quality index and education data deviation value, the applicability of each rule is calculated, and fuzzy reasoning is performed;

[0032] Convert fuzzy outputs into concrete predictive power values.

[0033] In a preferred embodiment, S6: the prediction ability of the model in different regions and educational environments is divided into high prediction ability, medium prediction ability and low prediction ability, specifically:

[0034] The obtained prediction ability value of the model is compared with the gradient standard threshold, the gradient standard threshold includes a first standard threshold and a second standard threshold, and the first standard threshold is less than the second standard threshold, and the prediction ability value of the model is compared with the first standard threshold and the second standard threshold respectively;

[0035] If the predictive ability value of the model is greater than the second standard threshold, the predictive ability of the model in different regions and educational environments is classified as high predictive ability; if the predictive ability value of the model is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, the predictive ability of the model in different regions and educational environments is classified as medium predictive ability; if the predictive ability value of the model is less than the first standard threshold, the predictive ability of the model in different regions and educational environments is classified as low predictive ability.

[0036] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0037] 1. This invention uses a variety of advanced technologies such as hierarchical analysis method, time series analysis method and fuzzy logic analysis to conduct in-depth analysis and processing of educational data in different regions and educational environments, and comprehensively considers key characteristic parameters such as the incremental index of teaching resource use and the abnormal index of teacher turnover rate. Through scientific evaluation and optimization, the generalization ability and prediction accuracy of the model in various regions have been significantly improved, which can effectively respond to the challenges brought by the imbalance of educational resources and differences in educational models, and ensure the reliability and accuracy of the evaluation results.

[0038] 2. The present invention provides differentiated solutions for different regions and educational environments by dynamically dividing and improving the prediction ability of the model. Specifically, by comparing with the gradient standard threshold, the high, medium, and low prediction capabilities of the model are classified and determined, and corresponding optimization adjustments are made based on the classification results, thereby achieving continuous improvement and optimization of the model performance. In this way, not only can the prediction inaccuracy of the model in certain areas be discovered and solved in a timely manner, but it can also provide scientific decision-making basis for education managers, optimize the allocation of educational resources, and ultimately improve the overall quality of education. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0040] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] For examples, see Figure 1 As shown, the education quality assessment method based on the machine learning model described in this embodiment includes the following steps:

[0043] S1: Determine the criteria for different regions and educational settings, and collect educational data from several regions and educational settings at multiple time points;

[0044] S2: Preprocess the education data from different sources and extract the characteristic parameters of the education data at multiple time points in different regions based on the evaluation criteria of education quality;

[0045] S3: Use the analytic hierarchy process to determine the changes in the characteristic parameters of each region at different time points, determine the weight assignment of each region, and calculate the overall education data quality index after weighted average calculation of the weight assignment of each region;

[0046] S4: Analyze the changing trends of educational data through time series analysis and evaluate the changes in the model's predictive performance when processing new data;

[0047] S5: After analyzing the overall education data quality index and the changes in the model's predictive performance when processing new data through fuzzy logic, the model's predictive ability in different regions and educational settings was evaluated;

[0048] S6: Based on the evaluation results, the predictive ability of the model in different regions and educational environments is divided into high predictive ability, medium predictive ability and low predictive ability, and the model is improved accordingly based on the division results.

[0049] Among them, in S1, the division criteria of different regions and educational environments are determined, and educational data at multiple time points are collected from several regions and educational backgrounds.

[0050] In order to effectively evaluate the quality of education in different regions and educational environments, it is necessary to first determine reasonable classification criteria, which can be based on a variety of factors such as geographical location, school type, and economic development level.

[0051] The division of regions includes: division of regions according to different administrative levels such as provinces, cities, and districts. In a province, division is made according to the nature and type of schools, such as public schools, private schools, key schools, ordinary schools, etc. Division is made according to the economic development level of the region, such as developed regions and underdeveloped regions.

[0052] According to the above classification criteria, education data at multiple time points are collected from several regions and educational backgrounds. Data collection can be divided into the following aspects:

[0053] Determine the time points for data collection: Select multiple representative time points, such as the midterm and final exam periods each year, or the beginning and end of each semester. For example, select the midterm and final exam periods of the past three years for data collection.

[0054] Collect education data: Collect data on education quality from various regions and schools, including but not limited to student performance, teacher evaluations, school facilities, teaching resources, etc.

[0055] Data source: Determine the specific source of the data, which may include statistical data from the education department, internal school records, data from third-party education evaluation agencies, etc.

[0056] Clean and organize the collected raw data, remove redundant and erroneous data, and ensure data quality. Classify the organized data according to the classification criteria and store them in the database for subsequent processing and analysis.

[0057] S2: Preprocess the education data from different sources and extract the characteristic parameters from the education data at multiple time points in different regions based on the evaluation criteria of education quality.

[0058] Preprocess educational data from different sources, check for missing values ​​in the data, and select appropriate methods to fill or delete them. Check and delete duplicate records in the data to ensure that each piece of data is unique. Unify the data formats from different sources, such as date format, numerical units, etc. Example: Unify all date formats to "YYYY-MM-DD" and convert all grade units to percentages. Ensure that the data types of all data fields are consistent and convert string values ​​to numeric types. Align data from different sources to ensure that the data is compared at the same time point and on the same scale.

[0059] Characteristic parameters are the core indicators for evaluating education quality. Based on the evaluation criteria for education quality, by extracting characteristic parameters from education data at multiple time points in different regions, we can better evaluate education quality.

[0060] Education quality evaluation standards are a series of indicators and criteria used to measure and evaluate the performance of the education system. These standards are usually formulated by education authorities, education research institutions or other relevant organizations, and are intended to comprehensively and objectively reflect the quality and effectiveness of the education system.

[0061] Education data for different regions, including provinces, cities, and districts, and data at multiple time points, including mid-term and final data for each semester.

[0062] Obtain characteristic parameters in the education data for each region and time point, including the incremental index of teaching resource utilization and the abnormal index of teacher turnover rate.

[0063] Among them, the method for obtaining the incremental index of teaching resource usage is:

[0064] Obtain the teaching resource usage data at the same time point t in each region, collect the independent variable data that affect the use of teaching resources, such as the number of students, the number of teachers, the size of the school, etc., ensure the integrity and consistency of the data, and perform necessary data cleaning and preprocessing;

[0065] According to the obtained teaching resource usage data, a multivariate linear regression model is selected, and the regression model is trained using historical data to fit the model parameters. The specific calculation expression of the multivariate linear regression model is: T t =(s0+s1X 1t +s2X 2t +...+s n X nt )+c; where T t represents the usage of teaching resources at time point t, X 1t , X 2t , ..., X ntare the influencing factors of the independent variables at time point t. These variables can be the number of students, the number of teachers, the size of the school, etc. s0, s1, s2, ..., s n is the regression coefficient, c is the error term, and the predicted teaching resource usage T at each time point is calculated. t ,

[0066] The teaching resource usage in the time period t+1 and the teaching resource usage in the time period t are analyzed, and the teaching resource usage increment index is calculated. The teaching resource usage increment index is obtained by calculating the ratio of the difference between the teaching resource usage in the time period t+1 and the teaching resource usage in the time period t and marking it as LB.

[0067] The incremental index of teaching resource usage can reflect the changing trend of teaching resource usage. By calculating the incremental index of each time period, it is possible to identify whether resource usage is increasing or decreasing, thereby helping education administrators understand the dynamic changes in resource allocation.

[0068] The incremental index can help education administrators optimize the allocation of teaching resources. According to the changes in the incremental index, education administrators can evaluate the effectiveness of current resource allocation and make corresponding adjustments to ensure that resources are optimally allocated.

[0069] The incremental index can be used as an important indicator for performance evaluation. By calculating the incremental index, the efficiency of resource utilization in different regions and time points can be evaluated, and then performance evaluation and improvement measures can be carried out.

[0070] The incremental index helps predict future resource needs. By analyzing historical data and the incremental index, education administrators can predict future demand for teaching resources and make plans in advance to avoid resource shortages or waste.

[0071] The method for obtaining the abnormal index of teacher turnover rate is as follows:

[0072] Collect teacher turnover rate data at various time points and regions, clean and standardize the data to ensure data integrity and consistency, and construct a prior distribution P(μ,σ 2 ), μ is the posterior mean, which represents the expected value of the teacher turnover rate, and σ is the posterior standard deviation, which represents the standard deviation of the teacher turnover rate;

[0073] According to historical data, the likelihood function of the turnover rate is established to calculate the probability distribution of the turnover rate under given data conditions. The specific calculation expression is: In the formula, P(R|μ,σ 2 ) is the probability distribution of turnover rate, R is the actual observed teacher turnover rate, and the posterior distribution is calculated using Bayesian theorem. The calculation expression is: P(μ,σ2 ∣R)∝P(R∣μ,σ 2 )P(μ,σ 2 ); P(μ,σ 2 |R) is the probability of abnormal situation under the condition of turnover rate R. The abnormal situation of teacher turnover rate is evaluated according to the posterior distribution. By calculating the posterior probability, the abnormal points are identified and measured, and the abnormal index of teacher turnover rate is calculated. The specific calculation expression is: Where BC is the abnormal index of teacher turnover rate.

[0074] The larger the abnormal index of teacher turnover rate is, the more problems there are in the education management of the region, specifically:

[0075] Low job satisfaction among teachers: High turnover rates usually reflect teachers’ dissatisfaction with their working environment, salary, benefits, career development, etc.

[0076] Insufficient management mechanism: The high teacher turnover rate may be due to an insufficiency of educational management mechanisms, such as the lack of effective incentive mechanisms and teacher training programs.

[0077] Improper resource allocation: High turnover rates may also indicate that there are problems with resource allocation in the school or education department, such as uneven distribution of teaching staff and excessive workload.

[0078] S3: The analytic hierarchy process is used to determine the changes in the characteristic parameters of each region at different time points, determine the weight assignment of each region, and calculate the overall education data quality index by performing weighted average calculation on the weight assignments of each region.

[0079] The hierarchical model is constructed to represent the goals, criteria and plans in three levels: goal level, criterion level and plan level. Goal level: overall education data quality index. Criterion level: teaching resource usage increase index, teacher turnover abnormality index. Plan level: data from different regions and time points. Judgment matrices are constructed for the criterion level and plan level respectively, and the relative importance of each element is determined by pairwise comparison.

[0080] Determine the importance between the increasing index of teaching resource use and the abnormal index of teacher turnover rate. If the importance of the increasing index of teaching resource use and the abnormal index of teacher turnover rate are equal, the judgment matrix of the criterion layer is: Normalize the matrix A and calculate the eigenvector W of the matrix A ;

[0081] For each region, the judgment matrix of the scheme layer is constructed based on the incremental index of teaching resource use and the abnormal index of teacher turnover rate, which are represented as B1 and B2 respectively. Each judgment matrix B1 and B is normalized to calculate the eigenvector W B1 and W B2 ;

[0082] Combine the weight vectors of the criterion layer and the solution layer to calculate the comprehensive weight of each region, W 总 =W A *(W B1 , W B2 );W 总 As the comprehensive weight of each region, the weight values ​​of each region are assigned and weighted averaged to obtain the overall education data quality index.

[0083] It should be noted that the normalization method of the judgment matrix is ​​a well-known technique and will not be described in detail here.

[0084] In the present invention, the changes in the incremental index of teaching resource use and the abnormal index of teacher turnover rate in each region at different time points are judged, the weight of each region is determined, and the overall education data quality index is calculated. The specific steps include constructing a hierarchical model, constructing a judgment matrix, calculating eigenvectors and eigenvalues, consistency testing, calculating comprehensive weights, and weighted average calculation. This method provides a scientific and systematic decision-making basis for education managers.

[0085] S4: Analyze the changing trends of educational data through time series analysis and evaluate the changes in the model's predictive performance when processing new data.

[0086] Collect historical education data, including student performance, teacher evaluations, school facilities, and teaching resources, establish a time series for the collected data, clean and standardize the data, and ensure the integrity and consistency of the data.

[0087] Decompose the time series data into trend, seasonal and random components, and select the moving average time series model ARIMA. The calculation expression of the model is: ec k =h+(v1G k-1 +v2G k-2 +...+v p G k-p +m1ec k-1 +m2ec k-2 +...+m p ec k-p );where ec k is the observed value at time point k, h is a constant term, which represents the average level of the sequence, v1, v2, ..., v p are the autoregressive coefficients, m1, m2, ..., m p is the moving average coefficient, ec k-1 ,ec k-2 ,...,ec k-p is the historical data value, G k-1 , Gk-2 , ..., G k-p is the error term;

[0088] Use historical data to train the selected time series model, fit the model parameters, split the data set into training set and test set, verify the model's predictive performance by predicting the test set data, use the trained model to predict future education data, compare and analyze the actual data and predicted data, calculate the standard deviation between the actual data and the predicted data, record it as the education data deviation value, and use the education data deviation value to evaluate the change in the model's predictive performance when processing new data.

[0089] The obtained education data deviation value is compared with the reference threshold of the education data deviation value under the standard state set in the historical data. If the education data deviation value is greater than or equal to the reference threshold of the education data deviation value under the standard state set in the historical data, it means that the prediction performance of the model when processing new data is poor, and a model prediction performance abnormality signal is generated at this time; if the education data deviation value is less than the reference threshold of the education data deviation value under the standard state set in the historical data, it means that the prediction performance of the model when processing new data is good, and a model prediction performance normal signal is generated at this time.

[0090] S5: After analyzing the overall education data quality index and the changes in the model’s predictive performance when processing new data through fuzzy logic, the model’s predictive ability in different regions and educational settings was evaluated.

[0091] The overall education data quality index and education data deviation value are used as input items, and the model's predictive ability (high, medium, and low) is used as the output item;

[0092] Define fuzzy sets for each input and output variable, and select membership functions; define fuzzy sets for the overall education data quality index: low, medium, and high; define fuzzy sets for the education data deviation value: small, medium, and large; define fuzzy sets for the predictive ability of the model: low, medium, and high.

[0093] Construct fuzzy rules to map input variables to output variables. If the overall education data quality index is high and the education data deviation value is small, the model's predictive ability is high. If the overall education data quality index is medium and the education data deviation value is medium, the model's predictive ability is medium. If the overall education data quality index is low and the education data deviation value is large, the model's predictive ability is low.

[0094] Using Mamdan i reasoning, the membership of the input variables is applied to the fuzzy rules, the fuzzy output is calculated for the given overall education data quality index and education data deviation value, the applicability of each rule is calculated, and fuzzy reasoning is performed.

[0095] Convert the fuzzy output into a specific predictive power value using a defuzzification method (e.g., centroid method, maximum membership method).

[0096] S6: Based on the evaluation results, the predictive ability of the model in different regions and educational environments is divided into high predictive ability, medium predictive ability and low predictive ability, and the model is improved accordingly based on the division results.

[0097] Compare the obtained prediction ability value of the model with the gradient standard threshold, the gradient standard threshold in the present application includes a first standard threshold and a second standard threshold, and the first standard threshold is less than the second standard threshold, and compare the prediction ability value of the model with the first standard threshold and the second standard threshold respectively;

[0098] If the model's predictive ability value is greater than the second standard threshold, the model's predictive ability in different regions and educational environments is classified as high predictive ability; the existing model parameters and configurations are maintained, and further optimization can be performed to maintain or improve performance.

[0099] If the predictive ability value of the model is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, the predictive ability of the model in different regions and educational environments is classified as medium predictive ability; analyze the shortcomings of the model in the medium predictive ability area, and perform targeted optimization and adjustment.

[0100] If the model's predictive ability value is less than the first standard threshold, the model's predictive ability in different regions and educational settings is classified as low predictive ability. A comprehensive analysis of the model's deficiencies may require parameter adjustments, algorithm improvements, or data set expansion to improve the model's predictive ability.

[0101] In the present invention, by comparing the prediction ability value of the model with the defined gradient standard threshold, the prediction ability of the model in different regions and educational environments can be effectively divided into three levels: high, medium, and low. According to the classification results, the model can be improved accordingly to improve the overall prediction performance. This method provides a scientific and systematic decision-making basis for educational managers, which helps to continuously optimize the performance of educational models.

[0102] In this embodiment, by determining the division criteria of different regions and educational environments, educational data at multiple time points are collected from several regions and educational backgrounds; educational data from different sources are preprocessed, and characteristic parameters in educational data at multiple time points in different regions are extracted based on the evaluation criteria for educational quality; the changes in characteristic parameters at different time points in each region are judged by the hierarchical analysis method, the weight assignment of each region is determined, and the weight assignment of each region is weighted averaged to calculate the overall education data quality index; the change trend of education data is analyzed by the time series analysis method, and the change in the prediction performance of the model when processing new data is evaluated; after analyzing the changes in the overall education data quality index and the prediction performance of the model when processing new data by fuzzy logic, the prediction ability of the model in different regions and educational environments is evaluated; according to the evaluation results, the prediction ability of the model in different regions and educational environments is divided into high prediction ability, medium prediction ability and low prediction ability, and the model is improved accordingly according to the division results. The quality evaluation of education data and the accuracy of the prediction model are effectively improved. Through meticulous classification and improvement measures, we have achieved precise optimization of the model's prediction capabilities in different regions and educational environments, ensuring the rational allocation of educational resources and the continuous improvement of teaching quality.

[0103] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0104] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0105] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0106] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0107] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for evaluating education quality based on a machine learning model, characterized in that: The steps include: S1: Determine the criteria for different regions and educational settings, and collect educational data from several regions and educational settings at multiple time points; S2: Preprocess the education data from different sources and extract the characteristic parameters of the education data at multiple time points in different regions based on the evaluation criteria of education quality; S3: Use the analytic hierarchy process to determine the changes in the characteristic parameters of each region at different time points, determine the weight assignment of each region, and calculate the overall education data quality index after weighted average calculation of the weight assignment of each region; S4: Analyze the changing trends of educational data through time series analysis and evaluate the changes in the model's predictive performance when processing new data; S5: After analyzing the overall education data quality index and the changes in the model's predictive performance when processing new data through fuzzy logic, the model's predictive ability in different regions and educational settings was evaluated; S6: Based on the evaluation results, the predictive ability of the model in different regions and educational environments is divided into high predictive ability, medium predictive ability and low predictive ability, and the model is improved accordingly based on the division results.

2. The method for evaluating education quality based on a machine learning model according to claim 1, characterized in that: In S2, characteristic parameters in the education data of each region and time point are obtained, including the incremental index of teaching resource utilization and the abnormal index of teacher turnover rate.

3. The method for evaluating education quality based on a machine learning model according to claim 2, characterized in that: in, The method for obtaining the incremental index of teaching resource usage is as follows: Obtain the teaching resource usage data at the same time point t in each region, collect the independent variable data that affect the use of teaching resources, and ensure the integrity and consistency of the data; According to the obtained teaching resource usage data, a multivariate linear regression model is selected, and the regression model is trained using historical data to fit the model parameters. The specific calculation expression of the multivariate linear regression model is: T t =(s0+s1X 1t +s2X 2t +...+s n X nt )+c; where T t represents the usage of teaching resources at time point t, X 1t , X 2t , ..., X nt are the influencing factors of the independent variable at time point t, s0, s1, s2, ..., s n is the regression coefficient, c is the error term, and the predicted teaching resource usage T at each time point is calculated. t , The teaching resource usage in the time period t+1 and the teaching resource usage in the time period t are analyzed, and the teaching resource usage increment index is calculated. The teaching resource usage increment index is obtained by calculating the ratio of the difference between the teaching resource usage in the time period t+1 and the teaching resource usage in the time period t and marking it as LB.

4. The method for evaluating education quality based on a machine learning model according to claim 3, characterized in that: The method for obtaining the abnormal index of teacher turnover rate is as follows: Collect teacher turnover rate data at various time points and regions, clean and standardize the data, and construct the prior distribution P(μ,σ 2 ), μ is the posterior mean, which represents the expected value of the teacher turnover rate, and σ is the posterior standard deviation, which represents the standard deviation of the teacher turnover rate; According to historical data, the likelihood function of the turnover rate is established to calculate the probability distribution of the turnover rate under given data conditions. The specific calculation expression is: In the formula, P(R|μ,σ 2 ) is the probability distribution of turnover rate, R is the actual observed teacher turnover rate, and the posterior distribution is calculated using Bayesian theorem. The calculation expression is: P(μ,σ 2 ∣R)∝P(R∣μ,σ 2 )P(μ,σ 2 ); P(μ,σ 2 |R) is the probability of abnormal situation under the condition of turnover rate R. The abnormal situation of teacher turnover rate is evaluated according to the posterior distribution. By calculating the posterior probability, the abnormal points are identified and measured, and the abnormal index of teacher turnover rate is calculated. The specific calculation expression is: Where BC is the abnormal index of teacher turnover rate.

5. The method for evaluating education quality based on a machine learning model according to claim 4, characterized in that: In S3, a hierarchical model is constructed to represent the goals, criteria and plans hierarchically, which are divided into three levels: goal level, criterion level and plan level; the goal level is the overall education data quality index, the criterion level is the incremental index of teaching resource use and the abnormal index of teacher turnover rate, and the plan level is the data of different regions and time points. Judgment matrices are constructed for the criterion level and the plan level respectively to compare and determine the importance between the incremental index of teaching resource use and the abnormal index of teacher turnover rate. The criterion level matrix A is normalized and the eigenvector W of the matrix is ​​calculated. A ; For each region, the judgment matrix of the scheme layer is constructed based on the incremental index of teaching resource use and the abnormal index of teacher turnover rate, which are represented as B1 and B2 respectively. Each judgment matrix B1 and B is normalized to calculate the eigenvector W B1 and W B2 ; Combine the weight vectors of the criterion layer and the solution layer to calculate the comprehensive weight of each region, W 总 =W A *(W B1 , W B2 );W 总 As the comprehensive weight of each region, the weight values ​​of each region are assigned and weighted averaged to obtain the overall education data quality index.

6. The method for evaluating education quality based on a machine learning model according to claim 5, characterized in that: In S4, the changing trend of education data is analyzed by time series analysis, specifically: Collect historical education data, including student grades, teacher evaluations, school facilities, and teaching resources, and build a time series with the collected data. Decompose the time series data into trend, seasonal, and random components, and select the moving average time series model ARIMA. The calculation expression of the model is: ec k =h(v1G k-1 +v2g k-2 +...+v p G k-p +m1ec k-1 +m2ec k-2 +...+m p ec k-p );where ec k is the observed value at time point k, h is a constant term, which represents the average level of the sequence, v1, v2, ..., v p are the autoregressive coefficients, m1, m2, ..., m p is the moving average coefficient, ec k-1 ,ec k-2 ,...,ec k-p is the historical data value, G k-1 , G k-2 ,·..,G k-p is the error term; Use historical data to train the selected time series model, fit the model parameters, split the data set into training set and test set, verify the model's predictive performance by predicting the test set data, use the trained model to predict future education data, compare and analyze the actual data and predicted data, calculate the standard deviation between the actual data and the predicted data, record it as the education data deviation value, and use the education data deviation value to evaluate the change in the model's predictive performance when processing new data.

7. The method for evaluating education quality based on a machine learning model according to claim 6, characterized in that: The obtained education data deviation value is compared with the reference threshold of the education data deviation value under the standard state set in the historical data. If the education data deviation value is greater than or equal to the reference threshold of the education data deviation value under the standard state set in the historical data, an abnormal model prediction performance signal is generated; if the education data deviation value is less than the reference threshold of the education data deviation value under the standard state set in the historical data, a normal model prediction performance signal is generated.

8. The method for evaluating education quality based on a machine learning model according to claim 7, characterized in that: In S5, the overall education data quality index and the education data deviation value are used as input items, and the prediction ability of the model is used as the output item; Define fuzzy sets for each input and output variable and choose membership functions; Construct fuzzy rules to map input variables to output variables; Using Mamdani reasoning, the membership of input variables is applied to fuzzy rules, the fuzzy output is calculated for a given overall education data quality index and education data deviation value, the applicability of each rule is calculated, and fuzzy reasoning is performed; Convert fuzzy outputs into concrete predictive power values.

9. The method for evaluating the quality of education based on a machine learning model according to claim 8, characterized in that: S6: dividing the prediction ability of the model in different regions and educational environments into high prediction ability, medium prediction ability and low prediction ability, specifically: The obtained prediction ability value of the model is compared with the gradient standard threshold, the gradient standard threshold includes a first standard threshold and a second standard threshold, and the first standard threshold is less than the second standard threshold, and the prediction ability value of the model is compared with the first standard threshold and the second standard threshold respectively; If the predictive ability value of the model is greater than the second standard threshold, the predictive ability of the model in different regions and educational environments is classified as high predictive ability; if the predictive ability value of the model is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, the predictive ability of the model in different regions and educational environments is classified as medium predictive ability; if the predictive ability value of the model is less than the first standard threshold, the predictive ability of the model in different regions and educational environments is classified as low predictive ability.

Images