Teaching score prediction method and device based on stacking algorithm and electronic equipment

Abstract

The application provides a teaching score prediction method and device based on a stacking algorithm and electronic equipment. The method comprises: obtaining student evaluation data for data preprocessing and feature extraction, and obtaining a sample data set; training a plurality of regression models and parameter adjustment processing based on the sample data set, and selecting n target prediction models from the plurality of regression models according to a set model prediction evaluation index ranking; wherein n is greater than or equal to 2; selecting a stacking model integration strategy, constructing n base learners of a first layer of a stacking model with the target prediction models, and taking each target prediction model as a meta-learner of a second layer of the stacking model; selecting a target integrated model, and determining a teaching score prediction model by comprehensively considering the target integrated model and a regression model with the optimal set model prediction evaluation index; and determining an output result of the teaching score prediction model as a teaching score prediction result. The application evaluates the teaching quality of teachers by comprehensively considering a plurality of evaluation indexes and regression models, and the evaluation result has high accuracy.

Description

Technical Field

[0001] This application relates to the field of educational technology, and in particular to a teaching score prediction method, device, and electronic device based on the Stacking algorithm. Background Technology

[0002] In today's era of big data, data mining is widely used in education, and educational data mining (EDM) has become a research hotspot in recent years. EDM uses methods such as clustering, regression, classification, and association rules to uncover the potential value of educational data, applying it to student behavior analysis, student learning outcome prediction, and teacher teaching quality prediction. Related technologies provide solutions for predicting student academic performance using principal component analysis, linear regression, and classification algorithms, thereby using the predicted student academic performance results as an indicator for evaluating teacher teaching quality.

[0003] In the process of implementing the embodiments of this application, at least the following technical problems were found in the related art:

[0004] The research mainly focuses on student performance. However, the quality of teacher evaluation data varies due to students' perfunctory evaluations, and the original evaluation results cannot truly reflect the teacher's teaching ability. Summary of the Invention

[0005] This application provides a teaching evaluation prediction method, device, and electronic device based on the Stacking algorithm to address the problem that the quality of evaluation data for teachers' teaching quality assessment is inconsistent, and the original evaluation results cannot truly reflect the teachers' teaching abilities.

[0006] In a first aspect, embodiments of this application provide a teaching score prediction method based on the Stacking algorithm, including:

[0007] Obtain student evaluation data for evaluating teachers' teaching quality; perform data preprocessing and feature extraction on the student evaluation data to obtain a sample dataset;

[0008] Multiple regression models are trained based on the sample dataset. The parameters of each regression model are adjusted using a grid optimization algorithm. Then, n target prediction models are selected from the multiple regression models according to the set model prediction evaluation index, where n≥2.

[0009] By selecting the Stacking model ensemble strategy, n base learners of the first layer of the Stacking model are constructed using the target prediction model, and each target prediction model is used as the meta learner of the second layer of the Stacking model to obtain n candidate ensemble models.

[0010] Train the n candidate ensemble models based on the sample dataset, select the model with the best prediction evaluation index as the target ensemble model, and combine the target ensemble model and the regression model with the best prediction evaluation index to determine the teaching score prediction model.

[0011] The output of the teaching score prediction model is defined as the teaching score prediction result.

[0012] In one possible implementation, determining the teaching score prediction model by integrating the target ensemble model and the regression model that optimally predicts the evaluation indicators includes:

[0013] The product of the target integrated model and the first weight is added to the product of the regression model that best predicts the evaluation index and the second weight to obtain the teaching score prediction model; wherein the first weight is greater than the second weight.

[0014] In one possible implementation, the set model prediction evaluation index is the mean squared error.

[0015] In one possible implementation, the step of performing data preprocessing and feature extraction on the student evaluation data to obtain a sample dataset includes:

[0016] After removing outliers and filling missing values ​​in the student evaluation data, the students' ratings and comments on the teachers are extracted as feature extraction results.

[0017] Establish reliability feature indicators based on feature extraction results;

[0018] The student evaluation data is filtered based on the reliability feature indicators to obtain a sample dataset.

[0019] In one possible implementation, the reliability feature indicators include at least three items;

[0020] The step of filtering the student evaluation data according to the reliability feature index to obtain a sample dataset includes:

[0021] The correlation coefficient method is used to calculate the correlation between various reliability characteristic indicators;

[0022] Based on the correlation calculation results, when there are reliability feature indicators with a correlation greater than the set value, one of the reliability feature indicators is retained, and the retained reliability feature indicator and the reliability feature indicators with a correlation less than the set value are clustered together;

[0023] The student ID-based evaluation data was selected based on the clustering results and used as a sample dataset for predicting evaluation scores.

[0024] In one possible implementation, the reliability features include multiple indicators such as: range of fluctuation, discrimination variance, adjacent difference variance, maximum ratio of identical scores, information content of comments, and proportion of different comments;

[0025] Wherein, the fluctuation range is the difference between the highest and lowest scores given in the student evaluation data of the same student;

[0026] The discrimination variance is the average of the squared differences between the scores given to each course and the average scores given to all courses in the student evaluation data of the same student.

[0027] The variance of adjacent differences is the variance of the difference between adjacent scores in the student evaluation data of the same student after the scores are sorted.

[0028] The maximum identical score ratio is the ratio of the number of the highest identical scores among the student evaluation data for the same student to the total number of scores.

[0029] The information content of the comments is the average number of effective words in the student evaluation data for the same student; where the effective word count is the total number of words corresponding to the effective words that are successfully matched with the preset evaluation dictionary.

[0030] The ratio of different comments to the total number of comments given by students in the same student's evaluation data.

[0031] In one possible implementation, obtaining student evaluation data for evaluating teacher teaching quality includes:

[0032] Obtain student information forms, teacher information forms, and teaching evaluation forms;

[0033] The student information table, the teacher information table, and the teaching evaluation table are merged to obtain student evaluation data.

[0034] Secondly, embodiments of this application provide a teaching score prediction device based on the Stacking algorithm, comprising:

[0035] The acquisition module is used to acquire student evaluation data for evaluating teachers' teaching quality, perform data preprocessing and feature extraction on the student evaluation data, and acquire a sample dataset.

[0036] The selection module is used to train multiple regression models based on the sample dataset, adjust the parameters of each regression model using a grid optimization algorithm, and select n target prediction models from the multiple regression models according to the set model prediction evaluation index; where n≥2;

[0037] The model building module is used to select a Stacking model ensemble strategy, construct n base learners for the first layer of the Stacking model using the target prediction model, and use each target prediction model as a meta-learner for the second layer of the Stacking model to obtain n candidate ensemble models.

[0038] Train the n candidate ensemble models based on the sample dataset, select the candidate ensemble model with the best prediction evaluation index as the target ensemble model, and combine the target ensemble model and the regression model with the best prediction evaluation index to determine the teaching score prediction model.

[0039] The determination module is used to determine the output of the teaching score prediction model as the teaching score prediction result.

[0040] In one possible implementation, the model building module is specifically used for:

[0041] The product of the target integrated model and the first weight is added to the product of the regression model that best predicts the evaluation index and the second weight to obtain the teaching score prediction model; wherein the first weight is greater than the second weight.

[0042] In one possible implementation, the set model prediction evaluation index is the mean squared error.

[0043] In one possible implementation, the plurality of regression models includes three or more of the following: logistic regression, lasso regression, linear regression, ridge regression, kernel ridge regression, elastic network, support vector machine, K-nearest neighbor regression, decision tree regression, extreme random tree, random forest, AdaBoost, GBRT, XGBoost and LightGBM.

[0044] In one possible implementation, the acquisition module is specifically used for:

[0045] After removing outliers and filling missing values ​​in the student evaluation data, the students' ratings and comments on the teachers are extracted as feature extraction results.

[0046] Establish reliability feature indicators based on feature extraction results;

[0047] The student evaluation data is filtered based on the reliability feature indicators to obtain a sample dataset.

[0048] In one possible implementation, the reliability feature indicators include at least three items;

[0049] The acquisition module is specifically used for:

[0050] The correlation coefficient method is used to calculate the correlation between various reliability characteristic indicators;

[0051] Based on the correlation calculation results, when there are reliability feature indicators with a correlation greater than the set value, one of the reliability feature indicators is retained, and the retained reliability feature indicator and the reliability feature indicators with a correlation less than the set value are clustered together;

[0052] The student ID-based evaluation data was selected based on the clustering results and used as a sample dataset for predicting evaluation scores.

[0053] In one possible implementation, the reliability features include multiple indicators such as: range of fluctuation, discrimination variance, adjacent difference variance, maximum ratio of identical scores, information content of comments, and proportion of different comments;

[0054] Wherein, the fluctuation range is the difference between the highest and lowest scores given in the student evaluation data of the same student;

[0055] The discrimination variance is the average of the squared differences between the scores given to each course and the average scores given to all courses in the student evaluation data of the same student.

[0056] The variance of adjacent differences is the variance of the difference between adjacent scores in the student evaluation data of the same student after the scores are sorted.

[0057] The maximum identical score ratio is the ratio of the number of the highest identical scores among the student evaluation data for the same student to the total number of scores.

[0058] The information content of the comments is the average number of effective words in the student evaluation data for the same student; where the effective word count is the total number of words corresponding to the effective words that are successfully matched with the preset evaluation dictionary.

[0059] The ratio of different comments to the total number of comments given by students in the same student's evaluation data.

[0060] In one possible implementation, the acquisition module is specifically used for:

[0061] Obtain student information forms, teacher information forms, and teaching evaluation forms;

[0062] The student information table, the teacher information table, and the teaching evaluation table are merged to obtain student evaluation data.

[0063] Thirdly, embodiments of this application provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method as described in the first aspect or any possible implementation of the first aspect above.

[0064] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the method as described in the first aspect or any possible implementation of the first aspect.

[0065] This application provides a teaching score prediction method, device, and electronic device based on the Stacking algorithm. By employing a Stacking model fusion algorithm to integrate multiple regression models to construct a teaching score prediction model, it integrates the advantages of various models, improving prediction accuracy and efficiency. Regarding teacher teaching quality, unlike relying solely on student grades, it utilizes student evaluation data containing multifaceted feedback to train regression and Stacking models. Student evaluation data makes teaching evaluation more scientific and fair; students' perspectives provide more diverse and comprehensive teaching feedback, improving prediction accuracy. This allows administrators to effectively manage teachers based on teaching score prediction results, avoiding significant discrepancies between teacher teaching methods and students' needs and expectations. Simultaneously, it provides administrators with a reference for teacher promotion and demotion, contributing to improved educational quality and offering guidance for future student course selection and administrator decision-making. Attached Figure Description

[0066] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0067] Figure 1 This is a flowchart illustrating the implementation of a teaching score prediction method based on the Stacking algorithm according to an embodiment of this application.

[0068] Figure 2 This is a schematic diagram illustrating the basic principles of Stacking;

[0069] Figure 3 This is a flowchart illustrating the implementation of a teaching score prediction method based on the Stacking algorithm according to a specific embodiment of this application.

[0070] Figure 4This is a histogram of the mean squared errors of the target integration model, the teaching score prediction model, and the basic regression model used in the teaching score prediction model provided in a specific embodiment of this application.

[0071] Figure 5 This application provides a specific embodiment of the target value before and after Box-Cox transformation during the data preprocessing and feature extraction process of student evaluation data;

[0072] Figure 6 This is a schematic diagram illustrating the relationship between the contour coefficient and the k value according to a specific embodiment of this application;

[0073] Figure 7 This is a schematic diagram of the structure of a teaching score prediction device based on the Stacking algorithm provided in an embodiment of this application;

[0074] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0075] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0076] The terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion.

[0077] Unless otherwise stated, the term "multiple" means two or more. The character " / " indicates that the preceding and following objects are in an "or" relationship. For example, A / B means: A or B. The term "and / or" describes an association between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or, A and B.

[0078] The terms used in this application are for describing embodiments only and are not intended to limit the claims. As used in the description of embodiments and claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. Similarly, the term “and / or” as used herein means including one or more of the associated listed elements and all possible combinations thereof. Additionally, when used in this application, the terms “comprise” and its variations “comprises” and / or “comprising” refer to the presence of stated features, integrals, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof. Without further limitation, an element defined by the phrase “comprising an…” does not exclude the presence of additional identical elements in the process, method, or apparatus that includes said element.

[0079] In this application, each embodiment focuses on describing the differences from other embodiments, and similar or identical parts between embodiments can be referred to mutually. For methods, products, etc., disclosed in the embodiments, if they correspond to the method section disclosed in the embodiments, then the relevant parts can be referred to the description of the method section.

[0080] Unlike other related technologies that primarily evaluate teaching quality based on student performance, this application shifts the focus to the teachers themselves. It aims to predict teaching scores by integrating student evaluation data, providing targeted suggestions to teachers and offering administrators effective references for teacher promotions and demotions, thereby contributing to the overall improvement of education quality.

[0081] In practice, student evaluation data primarily includes two types: grades and student comments. Student evaluation data, derived directly from students, accurately reflects their needs and expectations regarding teaching. This helps teachers better understand students' needs, thereby adjusting teaching strategies and improving teaching methods.

[0082] Student comments typically include evaluations of the teaching content, methods, and pace, providing teachers with objective feedback on teaching effectiveness. This helps teachers gain a comprehensive understanding of their teaching, identify strengths and weaknesses, and make targeted improvements. Furthermore, student comments often contain suggestions and opinions, offering teachers new teaching ideas and methods. This helps stimulate teaching innovation and promotes educational reform and progress.

[0083] In the data extraction phase, this application further selected student evaluation data with high reliability and then filtered out evaluation data from the group of students who provided serious evaluations, providing a solid data foundation for subsequent data mining. Furthermore, the core of this application is a Stacking-based evaluation score prediction model. This model can comprehensively utilize the advantages of different models to achieve better predictive ability (e.g., achieving a lower mean squared error). This application not only provides a reference for future student course selection but also offers some guidance for administrators in making decisions.

[0084] To make the objectives, technical solutions, and advantages of this application clearer, the following description will be provided in conjunction with the accompanying drawings and specific embodiments.

[0085] Figure 1 This is a flowchart illustrating the implementation of a teaching score prediction method based on the Stacking algorithm according to an embodiment of this application. Figure 1 As shown, the method includes the following steps:

[0086] S101, Obtain student evaluation data for teacher teaching quality evaluation, perform data preprocessing and feature extraction on the student evaluation data, and obtain a sample dataset.

[0087] In actual evaluation processes, student evaluation data directly reflects teaching quality, primarily comprising two types: scores and comments. However, students may be influenced by various factors during the evaluation process, leading to missing or incomplete scores or comments. To accurately evaluate teaching quality, this data needs preprocessing to fill in missing values ​​and eliminate outliers, thereby making the data more complete and accurate. Various methods and techniques can be used for data preprocessing, such as interpolation, regression, and clustering. The specific method chosen depends on the actual data and evaluation requirements.

[0088] In addition, feature extraction is required for data such as ratings and comments to obtain data that is valuable for prediction. For example, text analysis techniques can be used to extract features such as keywords and sentiment from comments, or statistical analysis techniques can be used to extract indicators such as distribution characteristics and mean from ratings.

[0089] In summary, before conducting teaching quality evaluations based on student evaluation data, it is necessary to preprocess the data and extract features to ensure the accuracy and reliability of the evaluation results.

[0090] S102: Train multiple regression models based on the sample dataset, adjust the parameters of each regression model using a grid optimization algorithm, and select n target prediction models from the multiple regression models according to the set model prediction evaluation index. Where n≥2.

[0091] Figure 2 This is a schematic diagram illustrating the basic principle of Stacking. Stacking first builds multiple first-level learners of different types and uses them to obtain first-level prediction results. Then, based on these first-level prediction results, a second-level learner is built to obtain the final prediction result.

[0092] like Figure 2 As shown, taking a first-layer learner consisting of three learners as an example, the features x and labels y are first input into the first-layer learner, which learns independently. Then, the predictions from the three learners are stacked column-wise. The stacked predictions are used as the feature values ​​of the second-layer learner, and the original label values ​​are used as the label values. The feature values ​​and label values ​​are then input into the second-layer learner for fitting. The second-layer learner integrates the model results from the three learners in the first layer.

[0093] In this embodiment, the aim is to construct a teaching score prediction model using a Stacking model ensemble strategy. Therefore, based on the fundamental principles of Stacking, at least two target prediction models need to be selected from multiple regression models to construct multiple base learners for the first layer of the Stacking model. These base learners will serve as inputs to the second layer model to further optimize the model's prediction performance.

[0094] In the specific implementation process, firstly, the sample dataset is divided into a training set and a test set. Then, the regression model is trained and tested using the training set and the test set, respectively. Next, we rank each regression model according to its prediction evaluation metrics (such as mean squared error MSE, mean squared error RMSE, etc.) and select the regression model with the highest ranking and better prediction performance as the base learner.

[0095] To improve the model's prediction accuracy, we rank multiple regression models based on their prediction evaluation metrics and select the top-ranked model with superior prediction performance to construct the base learner. This ensures that the selected base learner has higher predictive power and generalization performance, thereby improving the overall prediction accuracy of the Stacking model.

[0096] S103, select the Stacking model ensemble strategy, construct n base learners for the first layer of the Stacking model using the target prediction model, and use each target prediction model as the second layer meta learner of the Stacking model to obtain n candidate ensemble models.

[0097] S104. Train n candidate ensemble models based on the sample dataset, select the model with the best predicted evaluation index as the target ensemble model, and combine the target ensemble model and the regression model with the best predicted evaluation index to determine the teaching score prediction model.

[0098] S105, determine the output of the teaching score prediction model as the teaching score prediction result.

[0099] In this embodiment, we employ a stacking model ensemble strategy to obtain the final ensemble model. To achieve this, we first construct n base learners for the first layer, each built based on the target prediction model. Then, we input the prediction results of these n base learners into the second-layer meta-learner. This process involves constructing the second-layer meta-learner using the regression models corresponding to the n base learners in the first layer, thereby obtaining n candidate ensemble models.

[0100] In the meta-learner, not only are the prediction results of n base learners integrated, but the model is further optimized and adjusted. This process results in high prediction accuracy and generalization ability, enabling the model to better adapt to various situations. Finally, we select the model with the optimal prediction evaluation metric as the objective ensemble model. This objective ensemble model not only considers the prediction results of the base learners but also further optimizes and adjusts the model, resulting in higher prediction accuracy.

[0101] At the same time, we also combined the goal integration model and the regression model that optimally predicted the evaluation indicators to determine the final teaching score prediction model. This prediction model can better reflect the characteristics and patterns of teaching scores and provide more accurate and reliable support for education assessment and decision-making.

[0102] In this embodiment, a teaching score prediction model is constructed by integrating multiple regression models using a Stacking model fusion algorithm. This integrates the advantages of various models, improving prediction accuracy and efficiency. Regarding teacher teaching quality, unlike relying solely on student grades, student evaluation data containing multifaceted feedback is used to train the regression and Stacking models. Student evaluation data makes teaching evaluation more scientific and fair; students' perspectives provide more diverse and comprehensive teaching feedback, improving prediction accuracy. This allows administrators to effectively manage teachers based on the teaching score prediction results, avoiding significant discrepancies between teacher teaching methods and students' needs and expectations. Simultaneously, the prediction of teacher teaching scores provides administrators with a reference for teacher promotion and demotion, contributing to improved educational quality. Furthermore, it offers valuable insights for future student course selection and administrator decision-making.

[0103] In different embodiments, the integrated target model and the regression model that sets the optimal evaluation index for determining the teaching score prediction model are used in different ways.

[0104] In one possible implementation, the teaching score prediction model is determined by integrating the integrated goal model and setting the optimal regression model for predicting evaluation indicators, including:

[0105] The teaching score prediction model is obtained by adding the product of the target integrated model and the first weight to the product of the regression model with the second weight that is set as the optimal predictive evaluation index; wherein the first weight is greater than the second weight.

[0106] First, by adding the product of the target integration model and the first weight to the product of the regression model with the second weight that is set as the optimal predictive evaluation index, the prediction results of the two models can be balanced to a certain extent, resulting in a more comprehensive and accurate teaching score prediction model.

[0107] Secondly, the first weight is greater than the second weight, which means that the prediction results of the goal-integrated model have a larger proportion in the final teaching score prediction model. This is because goal-integrated models usually consider more teaching score prediction results, thus more accurately reflecting students' learning and teachers' teaching quality.

[0108] In this embodiment, a teaching score prediction model is determined by combining the goal-integrated model and the regression model that optimally predicts the evaluation indicators from the predefined model. This model considers not only the prediction results of the goal-integrated model but also the prediction results of the regression model that optimally predicts the evaluation indicators from the predefined model. By assigning different weights to the prediction results of these two models and synthesizing them, a teaching score prediction model is obtained, which can more accurately predict teaching scores and provide schools and teachers with more objective and effective assessments and feedback. At the same time, this also provides strong support and assurance for further optimizing teaching methods and improving teaching quality.

[0109] In another possible implementation, the teaching score prediction model is determined by integrating the integrated goal model and setting the optimal regression model for predicting evaluation indicators, including:

[0110] The teaching score prediction model is determined by the difference between the prediction results of the target integration model and the regression model that predicts the optimal evaluation indicators.

[0111] When the difference is within the set error range, the teaching score prediction model is the regression model that best predicts the evaluation indicators using the target integrated model or the set model.

[0112] Otherwise, the product of the target integrated model and the first weight is added to the product of the regression model with the second weight that is set as the optimal predictive evaluation index, and this product is used as the teaching score prediction model; where the first weight is greater than the second weight.

[0113] First, the teaching score prediction model is determined based on the difference between the prediction results of the objective ensemble model and the regression model that best predicts the evaluation indicators. This difference reflects the degree of difference between the two models in predicting teaching scores. Then, when the difference is within a set error range, the regression model that best predicts the evaluation indicators, either the objective ensemble model or the regression model, is selected as the teaching score prediction model. This ensures the accuracy and stability of the model. However, if the difference exceeds the set error range, a weighted merging method is used to combine the prediction results of the two models to form the teaching score prediction model. This method better integrates the advantages of the two models and improves the accuracy of the prediction.

[0114] In this embodiment, by comprehensively considering the prediction results of the target integration model and the regression model that sets the optimal prediction evaluation index, a more accurate and stable teaching score prediction model can be determined.

[0115] In practical implementation, model prediction evaluation indicators include mean absolute error, mean absolute percentage error, mean squared error, mean squared error, logarithm of mean squared error, and median absolute error. In different embodiments, in step S102 of the above embodiment, n target prediction models can be selected from multiple regression models based on any one or more ranking results. These evaluation indicators each have their own characteristics and are suitable for different situations and needs.

[0116] In one possible implementation, when selecting n target prediction models using a model prediction evaluation index, the model prediction evaluation index is set as mean square error.

[0117] Specifically, the top n target prediction models are selected in ascending order of mean square error.

[0118] The specific number of n target prediction models needs to be determined based on the specific value of the mean square error of each model. Models with a mean square error lower than the set value are selected to avoid excessive mean square error that would reduce prediction performance.

[0119] In this embodiment, mean squared error (MSE) is a common metric for evaluating model predictions; it reflects the average difference between the model's predicted values ​​and the actual values. Choosing a model with a lower MSE implies choosing a model with better predictive performance.

[0120] In other possible implementations, when selecting n target prediction models using multiple model prediction evaluation indicators, the model prediction evaluation indicators are set to any two or more of the following: mean absolute error, mean absolute percentage error, mean square error, mean square error, logarithm of mean square error, and median absolute error.

[0121] Specifically, when selecting n target prediction models using m model prediction evaluation indicators, the top A prediction models are selected according to the order of the first indicator, ..., and the top B prediction models are selected according to the order of the mth indicator. From the A and B prediction models, the m recurring prediction models are determined as target models.

[0122] In one possible implementation, the multiple regression models selected in step S102 include three or more of the following: logistic regression, lasso regression, linear regression, ridge regression, kernel ridge regression, elastic network, support vector machine, k-nearest neighbor regression, decision tree regression, extreme random tree, random forest, AdaBoost, GBRT, XGBoost, and LightGBM. These models are all widely used regression models in the field of machine learning, each with its own characteristics and advantages.

[0123] To predict data more accurately, multiple regression models are typically trained based on the sample dataset, and grid optimization algorithms are used to tune the parameters of each model. Therefore, in different implementations, depending on the specific data prediction requirements and prediction efficiency needs, a target prediction model can be selected from the above three or more regression models to reduce the workload of model training and optimization while improving the model's prediction accuracy.

[0124] Figure 3 This is a schematic diagram of the construction process based on the Stacking integration model provided in a specific embodiment of this application. As shown in the figure, it includes the following steps:

[0125] Obtain the dataset;

[0126] The feature variables in the dataset are used as input, and 15 basic models are built for fitting and prediction. The basic models include logistic regression, lasso regression, linear regression, ridge regression, kernel ridge regression, elastic network, support vector machine, K nearest neighbor regression, decision tree regression, extreme random tree, random forest, AdaBoost, GBRT, XGBoost and LightGBM.

[0127] Four optimal models were selected for Stacking fusion; these four optimal models served as the first layer base learners in Stacking; the four optimal models were AdaBoost, Random Forest, XGBoost, and Elastic Network.

[0128] The feature variables in the dataset are used as inputs to four base learners.

[0129] The prediction results of the first base learner are used as input to the second meta learner, AdaBoost.

[0130] The Stacking model is weighted with the basic model to obtain a new model, which is then used as a teaching score prediction model.

[0131] The prediction results based on the improved Stacking model were used as the final teaching score prediction results.

[0132] In this specific embodiment, n target prediction models are selected by combining the prediction evaluation indicators of 15 regression models. By comparing the prediction results of multiple models, the merits and demerits of each model can be evaluated more accurately, and the model selection can be further optimized.

[0133] After adjusting the parameters of each regression model using a grid optimization algorithm, the mean squared errors of the 15 regression models are determined as shown in Table 1 below:

[0134] Table 1. Mean Squared Errors of 15 Regression Models

[0135]

[0136]

[0137] As shown in the table above, the mean squared errors (MSEs) of different regression models vary after parameter adjustment. The lowest MSE is 11.5919, and the highest is 16.4588. The four algorithms with the highest prediction performance are AdaBoost, Random Forest, XGBoost, and Elastic Regression Network (ERN). Therefore, AdaBoost, Random Forest, XGBoost, and ERN are used as the first-layer base learners in the ensemble model, and AdaBoost, Random Forest, XGBoost, and ERN are used as the second-layer meta-learners, respectively. After multiple experiments, AdaBoost was selected as the second-layer meta-learner because it has the lowest MSE.

[0138] After selecting the AdaBoost algorithm as the second-layer meta-learner, the mean squared error of the Stacking ensemble model was 11.5187, slightly lower than that of the AdaBoost regression model. The model with the smaller error, the AdaBoost regression model, and the ensemble model were weighted and summed to obtain the teaching score prediction model. Specifically, the weight of the ensemble model was set to 0.955, and the weight of the AdaBoost regression model was set to 0.005.

[0139] The improved Stacking model, also known as the teaching score prediction model, has a mean squared error of 11.4372, which is the smallest among all models. The mean squared errors of the goal integration model and the teaching score prediction model are shown in Table 2.

[0140] Table 2. Mean Squared Errors of the Goal Integration Model and the Teaching Score Prediction Model

[0141]

[0142] Figure 4 It is a histogram of the mean squared error of the basic regression model used in the goal integration model, the teaching score prediction model, and the teaching score prediction model.

[0143] Combined Table 2 and Figure 4 It can be seen that the teaching score prediction model derived from the solution provided in the embodiments of this application has high accuracy.

[0144] In one possible implementation, student evaluation data undergoes data preprocessing and feature extraction to obtain a sample dataset, including:

[0145] After removing outliers and filling in missing values ​​in the student evaluation data, the students' ratings and comments on the teachers are extracted as feature extraction results.

[0146] Establish reliability feature indicators based on feature extraction results;

[0147] Student evaluation data were filtered based on reliability indicators to obtain a sample dataset.

[0148] This process involves outlier removal and missing value imputation of student evaluation data to ensure the integrity and accuracy of the dataset. When removing outliers, the data distribution must be considered to avoid mistakenly deleting valid data. Optionally, missing value imputation can be performed using statistical methods, regression methods, or artificial intelligence-based algorithms for prediction.

[0149] Extracting students' ratings and comments for teachers as feature extraction results can represent multiple aspects of the ratings and evaluations. Based on the feature extraction results, reliability indicators can be established to determine the level of effort students put into making evaluations, and the reliability of the evaluation data can be determined according to the level of effort.

[0150] By filtering student evaluation data based on reliability indicators, a sample dataset can be obtained, which can exclude evaluation data with low reliability and retain evaluation data with high reliability.

[0151] In a specific embodiment, to analyze the rating and comment data in student evaluation data, the data features are transformed into numerical values. Ordered category data are assigned values ​​sequentially, while unordered category data undergoes one-hot encoding, followed by normalization and standardization to enhance data comparability. Since the target values ​​are not normally distributed, after multiple experiments, a Box-Cox transformation was chosen. The Box-Cox transformation is a mathematical transformation method that improves the normality and homogeneity of variance of the data, making the model more stable and reliable. The rating distribution before and after the transformation is shown below. Figure 5 .

[0152] In addition, the selected high-reliability evaluation data were split into training and test sets in a 7:3 ratio, and these sets were used as the sample datasets. The training set was used to train the model, and the test set was used to evaluate the model's performance and generalization ability. This approach better simulates real-world scenarios and improves the model's interpretability and generalization ability.

[0153] In this embodiment, data preprocessing is performed to make the data more standardized and accurate, avoiding adverse effects on subsequent analysis. After data preprocessing, student ratings and comments for teachers are extracted as feature extraction results, helping to better understand and analyze student evaluation data. After feature extraction, reliability indicators are established based on these features to measure the reliability and accuracy of student evaluation data, thereby better filtering out high-quality data. Student evaluation data is filtered according to the reliability indicators to obtain the final sample dataset, which helps to better understand the characteristics and patterns of student evaluation data, providing important support and reference for subsequent research and analysis.

[0154] In one possible implementation, the reliability feature indicators include at least three items;

[0155] Student evaluation data was filtered based on reliability indicators to obtain a sample dataset, including:

[0156] The correlation coefficient method is used to calculate the correlation between various reliability characteristic indicators;

[0157] Based on the correlation calculation results, when there are reliability feature indicators with a correlation greater than the set value, one of the reliability feature indicators is retained, and the retained reliability feature indicator and the reliability feature indicators with a correlation less than the set value are clustered together;

[0158] The student ID-based evaluation data was selected based on the clustering results and used as a sample dataset for predicting evaluation scores.

[0159] In practice, each rating process requires selecting reliability features based on relevance, retaining only one feature from several highly correlated features. To ensure the accuracy and reliability of the sample dataset, student evaluation data is filtered based on reliability feature indicators to obtain a representative sample dataset.

[0160] When calculating the correlation between various reliability feature indicators, the correlation coefficient method is used. This method effectively measures the correlation between indicators, facilitates feature selection, and helps identify features important to the model, thereby reducing the number of features, running time, and improving model accuracy. Based on the correlation calculation results, when a reliability feature indicator with a correlation greater than a set value appears, one of the reliability feature indicators is retained, while other reliability feature indicators with a correlation greater than the set value are removed. This avoids data redundancy and model overfitting, ensuring that the model can more accurately reflect the underlying patterns in the data.

[0161] Meanwhile, the retained reliability features and reliability features with a correlation of less than the set value are clustered. By grouping similar data points, the data points within the same group are more similar to each other, and the data points between different groups are as different as possible.

[0162] During the clustering process, students are grouped based on the similarity between reliability features, and the student evaluation data from the high-reliability groups are selected as sample datasets for predicting evaluation scores. These sample datasets will be used to train and validate the prediction model to evaluate student evaluation scores.

[0163] In this embodiment, the correlation coefficient method is used to calculate the correlation between various reliability feature indicators. When a reliability feature indicator has a correlation greater than a set value, one of the reliability feature indicators is retained for feature selection, avoiding data redundancy and model overfitting. The retained reliability feature indicators and the reliability feature indicators with correlation less than the set value are clustered, and the evaluation data of student IDs in the high reliability group of the clustering results are used as the sample dataset for predicting evaluation scores, thereby improving the accuracy of the prediction results.

[0164] In one possible implementation, reliability features include multiple indicators such as: range of fluctuation, discrimination variance, adjacent difference variance, maximum identical score ratio, comment information content, and proportion of different comments.

[0165] Among them, the fluctuation range is the difference between the highest and lowest scores given by students in the same student evaluation data; the larger the fluctuation range, the stronger the volatility of student scores.

[0166] The discrimination variance is the average of the squared differences between the scores given by students for each course and the average scores given for all courses in the student evaluation data for the same student; the larger the discrimination variance, the greater the student's discrimination of the courses.

[0167] The adjacent difference variance is the variance of the difference between adjacent scores in the student evaluation data of the same student after the scores are sorted; the larger the adjacent difference variance, the greater the difference in student scores.

[0168] The maximum identical score ratio is the ratio of the number of the highest identical scores among the student evaluation data for the same student to the total number of scores. The higher the maximum identical score ratio, the less seriously students take the evaluation.

[0169] The information content of the comments is the average number of effective words in the student evaluation data for the same student; where the effective word count is the total number of words corresponding to the effective words that are successfully matched with the preset evaluation dictionary; the higher the average number of effective words in the comments given by students, the more seriously the students take the evaluation activity.

[0170] The ratio of different comments is the ratio of the number of different comments given by students to the total number of comments given by students in the same student's evaluation data. The higher the ratio of different comments, the more thoughtful the students are, and the stronger the objectivity of their comments.

[0171] In one specific embodiment, the reliability characteristics include: range of fluctuation, variance of discrimination, variance of adjacent differences, ratio of maximum identical scores, information content of comments, and proportion of different comments. The correlation between the various reliability characteristics is calculated using the correlation coefficient method, as shown in the table below:

[0172]

[0173]

[0174] Based on the table above, it is easy to see that the correlation coefficients of the three features—adjacent difference variance, discrimination variance, and fluctuation range—are all greater than 0.7, indicating a strong correlation. Multiple comparative experiments revealed that deleting adjacent difference variance and fluctuation range yielded the best results; therefore, adjacent difference variance and fluctuation range were deleted, while discrimination variance was retained.

[0175] Among them, data with a discrimination index of 0 represents student evaluations made randomly without thought; therefore, data with a discrimination index of 0 are directly classified into the lowest reliability category. K-means clustering is used, based on... Figure 6 The relationship between the silhouette coefficient and the k-value indicates that dividing the remaining data into two categories yields the best results. The cluster centers of the two clusters are shown in Table 3.

[0176] Table 3 Cluster centers of clusters

[0177]

[0178] In the specific implementation of this embodiment, the evaluation data of a group of middle school student IDs with high reliability values ​​are selected based on the clustering results as the sample dataset for predicting evaluation scores. Specifically, this includes selecting the evaluation data of student IDs with label 1 in the clustering results as the dataset for predicting evaluation scores. The dataset has teacher features and course features, and the target value of the dataset is the evaluation score. The dataset attributes and their descriptions are shown in Table 4:

[0179] Table 4 Dataset Attributes and Descriptions

[0180]

[0181]

[0182] In one possible implementation, student evaluation data for evaluating teacher teaching quality is obtained, including:

[0183] Obtain student information forms, teacher information forms, and teaching evaluation forms;

[0184] The student information sheet, teacher information sheet, and teaching evaluation sheet are combined to obtain student evaluation data.

[0185] In the specific implementation process, after merging the student information table, teacher information table, and teaching evaluation table, the characteristics and descriptions of the resulting data table are shown in Table 4 above.

[0186] It should be understood that the sequence number of each step in the above embodiments does not imply 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 this application.

[0187] The following are device embodiments of this application. For details not described in detail, please refer to the corresponding method embodiments described above.

[0188] Figure 7 This is a schematic diagram of the structure of a teaching score prediction device based on the Stacking algorithm provided in an embodiment of this application, as shown below. Figure 7 As shown, for ease of explanation, only the parts related to the embodiments of this application are shown, such as... Figure 7 As shown, the device includes:

[0189] Module 701 is used to acquire student evaluation data for teacher teaching quality evaluation, perform data preprocessing and feature extraction on the student evaluation data, and acquire a sample dataset.

[0190] Module 702 is selected to train multiple regression models based on the sample dataset, adjust the parameters of each regression model using a grid optimization algorithm, and select n target prediction models from the multiple regression models according to the set model prediction evaluation index; where n≥2;

[0191] The model building module 703 is used to select the Stacking model ensemble strategy, build n base learners of the first layer of the Stacking model with the target prediction model, and use each target prediction model as the second layer meta learner of the Stacking model to obtain n candidate ensemble models.

[0192] Train n candidate ensemble models based on the sample dataset, select the candidate ensemble model with the best prediction evaluation index as the target ensemble model, and combine the target ensemble model and the regression model with the best prediction evaluation index to determine the teaching score prediction model.

[0193] Module 704 is used to determine the output of the teaching score prediction model as the teaching score prediction result.

[0194] In one possible implementation, the model building module 703 is specifically used for:

[0195] The teaching score prediction model is obtained by adding the product of the target integrated model and the first weight to the product of the regression model with the second weight that is set as the optimal predictive evaluation index; wherein the first weight is greater than the second weight.

[0196] In one possible implementation, the model prediction evaluation index is set as mean squared error.

[0197] In one possible implementation, multiple regression models include three or more of the following: logistic regression, lasso regression, linear regression, ridge regression, kernel ridge regression, elastic network, support vector machine, K-nearest neighbor regression, decision tree regression, extreme random tree, random forest, AdaBoost, GBRT, XGBoost, and LightGBM.

[0198] In one possible implementation, module 701 is specifically used for:

[0199] After removing outliers and filling in missing values ​​in the student evaluation data, the students' ratings and comments on the teachers are extracted as feature extraction results.

[0200] Establish reliability feature indicators based on feature extraction results;

[0201] Student evaluation data were filtered based on reliability indicators to obtain a sample dataset.

[0202] In one possible implementation, the reliability feature indicators include at least three items;

[0203] Module 701 is used specifically for:

[0204] The correlation coefficient method is used to calculate the correlation between various reliability characteristic indicators;

[0205] Based on the correlation calculation results, when there are reliability feature indicators with a correlation greater than the set value, one of the reliability feature indicators is retained, and the retained reliability feature indicator and the reliability feature indicators with a correlation less than the set value are clustered together;

[0206] The student ID-based evaluation data was selected based on the clustering results and used as a sample dataset for predicting evaluation scores.

[0207] In one possible implementation, reliability features include multiple indicators such as: range of fluctuation, discrimination variance, adjacent difference variance, maximum identical score ratio, comment information content, and proportion of different comments.

[0208] Among them, the fluctuation range is the difference between the highest and lowest scores given by students in the same student's evaluation data;

[0209] The variance of discrimination is the average of the squared differences between the scores given to each course and the average scores given to all courses in the student evaluation data of the same student.

[0210] The variance of adjacent differences is the variance of the difference between adjacent scores in the student evaluation data of the same student after the scores are sorted.

[0211] The maximum identical score ratio is the ratio of the number of the highest identical scores among the student evaluation data for the same student to the total number of scores.

[0212] The information content of the comments is the average number of effective words in the student evaluation data for the same student; where the effective word count is the total number of words corresponding to the effective words that are successfully matched with the preset evaluation dictionary.

[0213] The ratio of different comments to the total number of comments given by students in the same student's evaluation data.

[0214] In one possible implementation, module 701 is specifically used for:

[0215] Obtain student information forms, teacher information forms, and teaching evaluation forms;

[0216] The student information sheet, teacher information sheet, and teaching evaluation sheet are combined to obtain student evaluation data.

[0217] In this embodiment, a teaching score prediction model is constructed by integrating multiple regression models using a Stacking model fusion algorithm. This integrates the advantages of various models, improving prediction accuracy and efficiency. Regarding teacher teaching quality, unlike relying solely on student grades, student evaluation data containing multifaceted feedback is used to train the regression and Stacking models. Student evaluation data makes teaching evaluation more scientific and fair; students' perspectives provide more diverse and comprehensive teaching feedback, improving prediction accuracy. This allows administrators to effectively manage teachers based on the teaching score prediction results, avoiding significant discrepancies between teacher teaching methods and students' needs and expectations. Simultaneously, the prediction of teacher teaching scores provides administrators with a reference for teacher promotion and demotion, contributing to improved educational quality. Furthermore, it offers valuable insights for future student course selection and administrator decision-making.

[0218] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. For example... Figure 8 As shown, the electronic device 8 of this embodiment includes: a processor 80, a memory 81, and a computer program 82 stored in the memory 81 and executable on the processor 80. When the processor 80 executes the computer program 82, it implements the steps in the various embodiments of the teaching score prediction method based on the Stacking algorithm described above, for example... Figure 1 The steps shown. Alternatively, when the processor 80 executes the computer program 82, it implements the functions of each module / unit in the above-described device embodiments, for example... Figure 7 The functions of each module are shown.

[0219] For example, the computer program 82 can be divided into one or more modules / units, which are stored in the memory 81 and executed by the processor 80 to complete this application. The one or more modules / units can be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program 82 in the electronic device 8. For example, the computer program 82 can be divided into... Figure 7 The modules shown.

[0220] The electronic device 8 can be a desktop computer, laptop, handheld computer, or cloud server, etc. The electronic device 8 may include, but is not limited to, a processor 80 and a memory 81. Those skilled in the art will understand that... Figure 8 This is merely an example of electronic device 8 and does not constitute a limitation on electronic device 8. It may include more or fewer components than shown, or combine certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, etc.

[0221] The processor 80 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0222] The memory 81 can be an internal storage unit of the electronic device 8, such as a hard disk or memory. The memory 81 can also be an external storage device of the electronic device 8, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory 81 can include both internal and external storage units of the electronic device 8. The memory 81 is used to store the computer program and other programs and data required by the electronic device. The memory 81 can also be used to temporarily store data that has been output or will be output.

[0223] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0224] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0225] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0226] In the embodiments provided in this application, it should be understood that the disclosed devices / electronic devices and methods can be implemented in other ways. For example, the device / electronic device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings or direct couplings or communication connections may be through some interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0227] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0228] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0229] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above-described embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various embodiments of the teaching score prediction method based on the Stacking algorithm. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0230] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A teaching score prediction method based on the Stacking algorithm, characterized in that, include: Obtain student evaluation data for evaluating teachers' teaching quality, perform data preprocessing and feature extraction on the student evaluation data, and obtain a sample dataset; Multiple regression models are trained based on the sample dataset. The parameters of each regression model are adjusted using a grid optimization algorithm. Then, n target prediction models are selected from the multiple regression models according to the set model prediction evaluation index, where n≥2. By selecting the Stacking model ensemble strategy, n base learners of the first layer of the Stacking model are constructed using the target prediction model, and each target prediction model is used as the meta learner of the second layer of the Stacking model to obtain n candidate ensemble models. Train the n candidate ensemble models based on the sample dataset, select the candidate ensemble model with the best prediction evaluation index as the target ensemble model, and combine the target ensemble model and the regression model with the best prediction evaluation index to determine the teaching score prediction model. The output of the teaching score prediction model is defined as the teaching score prediction result; The step of determining the teaching score prediction model by integrating the target ensemble model and the regression model that optimizes the model prediction evaluation indicators includes: The product of the target integrated model and the first weight is added to the product of the regression model that best predicts the evaluation index and the second weight to obtain the teaching score prediction model; wherein the first weight is greater than the second weight.

2. The teaching score prediction method based on the Stacking algorithm according to claim 1, characterized in that, The set model prediction evaluation index is the mean square error.

3. The teaching score prediction method based on the Stacking algorithm according to claim 1, characterized in that, The process of preprocessing and feature extraction of the student evaluation data to obtain a sample dataset includes: After removing outliers and filling missing values ​​in the student evaluation data, the students' ratings and comments on the teachers are extracted as feature extraction results. Establish reliability feature indicators based on feature extraction results; The student evaluation data is filtered based on the reliability feature indicators to obtain a sample dataset.

4. The teaching score prediction method based on the Stacking algorithm according to claim 3, characterized in that, The reliability feature indicators include at least three items; The step of filtering the student evaluation data according to the reliability feature index to obtain a sample dataset includes: The correlation coefficient method is used to calculate the correlation between various reliability characteristic indicators; Based on the correlation calculation results, when there are reliability feature indicators with a correlation greater than the set value, one of the reliability feature indicators is retained, and the retained reliability feature indicator and the reliability feature indicator with a correlation less than the set value are clustered together; The student ID-based evaluation data was selected based on the clustering results and used as a sample dataset for predicting evaluation scores.

5. The teaching score prediction method based on the Stacking algorithm according to claim 4, characterized in that, The reliability characteristics include multiple items such as: range of fluctuation, discrimination variance, adjacent difference variance, maximum identical score ratio, comment information content, and proportion of different comments; Wherein, the fluctuation range is the difference between the highest and lowest scores given in the student evaluation data of the same student; The discrimination variance is the average of the squared differences between the scores given to each course and the average scores given to all courses in the student evaluation data of the same student. The variance of adjacent differences is the variance of the difference between adjacent scores in the student evaluation data of the same student after the scores are sorted. The maximum identical score ratio is the ratio of the number of the highest identical scores among the student evaluation data for the same student to the total number of scores. The information content of the comments is the average number of effective words in the student evaluation data for the same student; where the effective word count is the total number of words corresponding to the effective words that are successfully matched with the preset evaluation dictionary; the ratio of different comments is the ratio of the number of different comments given by the student to the total number of comments given by the student in the student evaluation data for the same student.

6. The teaching score prediction method based on the Stacking algorithm according to claim 1, characterized in that, The acquisition of student evaluation data for evaluating teacher teaching quality includes: Obtain student information forms, teacher information forms, and teaching evaluation forms; The student information table, the teacher information table, and the teaching evaluation table are merged to obtain student evaluation data.

7. A teaching score prediction device based on the Stacking algorithm, characterized in that, include: The acquisition module is used to acquire student evaluation data for teacher teaching quality evaluation, perform data preprocessing and feature extraction on the student evaluation data, and acquire a sample dataset. The selection module is used to train multiple regression models based on the sample dataset, adjust the parameters of each regression model using a grid optimization algorithm, and select n target prediction models from the multiple regression models according to the set model prediction evaluation index; where n≥2; The model building module is used to select a Stacking model ensemble strategy, construct n base learners for the first layer of the Stacking model using the target prediction model, and use each target prediction model as a meta-learner for the second layer of the Stacking model to obtain n candidate ensemble models. Train the n candidate ensemble models based on the sample dataset, select the candidate ensemble model with the best prediction evaluation index as the target ensemble model, and combine the target ensemble model and the regression model with the best prediction evaluation index to determine the teaching score prediction model. The determination module is used to determine the output of the teaching score prediction model as the teaching score prediction result; Specifically, the model building module is used for: The product of the target integrated model and the first weight is added to the product of the regression model that best predicts the evaluation index and the second weight to obtain the teaching score prediction model; wherein the first weight is greater than the second weight.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 6 above.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 6 above.

Images