A teaching quality comprehensive evaluation method based on teaching evaluation data credibility

Abstract

The application discloses a kind of based on teaching evaluation data credibility's comprehensive evaluation method of teaching quality, belong to data processing technical field, comprising: collection historical multi-period teaching evaluation original score and student academic record data, wash invalid data and construct associated data set;Based on data set, the homogeneity coefficient of each period evaluation index score of teacher is calculated, and the homogeneity correction factor is generated by combining homogeneity threshold value and effective questionnaire total number, and the original comprehensive score of teacher is corrected;Using historical data, the correlation coefficient of each evaluation index score and academic record is calculated, and the original weight of each index is dynamically adjusted by combining correlation coefficient threshold value and internal consistency reliability coefficient;Finally, according to the original score of the latest evaluation cycle, correction factor and the weight after correction, the final comprehensive evaluation score of teacher is calculated.The application introduces data credibility analysis and weight dynamic optimization, and improves the scientificity, fairness and accuracy of teaching evaluation.

Description

Technical Field

[0001] This invention belongs to the field of data processing technology, and in particular relates to a comprehensive evaluation method for teaching quality based on the credibility of teaching evaluation data. Background Technology

[0002] With the development of educational informatization, the collection and processing of teaching evaluation data has received increasing attention. How to extract effective information from multi-source data and improve the scientificity and fairness of evaluation results has become a hot topic in current teaching quality assessment research.

[0003] In existing technologies, teaching quality evaluation typically relies on student-completed questionnaire data, using a fixed-weighted summation method to calculate teachers' overall scores. Some methods incorporate student academic performance as a reference, attempting to establish a correlation between evaluation results and teaching effectiveness. However, existing methods largely focus on superficial data processing, lacking in-depth consideration of the reliability of the evaluation data itself; issues such as scoring consistency and sample validity remain unresolved.

[0004] Existing technologies have significant shortcomings in practical applications: First, they neglect the consistency within the evaluation data and fail to identify and correct for the homogeneity of scores from different teachers or different periods, making the evaluation results susceptible to extreme or arbitrary scoring. Second, the weighting of evaluation indicators lacks a dynamic adjustment mechanism and fails to optimize the weights by combining historical data and teaching effectiveness, making it difficult to reflect the actual contribution of the indicators. Third, there is a lack of systematic analysis of the correlation between evaluation data and academic performance, leading to a deviation between the evaluation results and the true level of teaching quality, affecting the objectivity and effectiveness of the evaluation. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a comprehensive evaluation method for teaching quality based on the reliability of teaching evaluation data, thus solving the aforementioned problems.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a comprehensive evaluation method for teaching quality based on the reliability of teaching evaluation data, comprising:

[0007] S1. Collect raw teaching evaluation scores and corresponding student academic performance data from multiple historical periods, clean up invalid data, and construct a related dataset containing teachers, evaluation indicators, and academic performance.

[0008] S2. Based on the associated dataset, obtain the homogenization coefficient of each teacher's evaluation indicator scores and the total number of valid questionnaires in each evaluation period. Based on the homogenization coefficient, the preset homogenization threshold and the total number of valid questionnaires, generate a homogenization correction factor. Use the homogenization correction factor to correct the teacher's original comprehensive score and output the homogenized individual score.

[0009] S3. Based on historical data from multiple periods in the associated dataset, obtain the score-grade correlation coefficient between the scores of each evaluation indicator and academic performance. Then, dynamically adjust the original weights of each evaluation indicator based on the score-grade correlation coefficient, the preset correlation coefficient threshold, and the internal consistency reliability coefficient of the evaluation indicator, and output the corrected indicator weights.

[0010] S4. Based on the original evaluation score of the latest evaluation cycle, the homogenization correction factor, and the corrected indicator weights, obtain and output the final comprehensive evaluation score of teaching quality for each teacher.

[0011] Based on the above technical solutions, the present invention also provides the following optional technical solutions:

[0012] Further technical solution: The specific steps of S1 are as follows:

[0013] Constructing the raw score matrix of teaching evaluation ,in, Indicates the first Period, No. Teacher, No. The original average score of each indicator For the first Valid questionnaire scores This indicates the total number of valid questionnaires;

[0014] Constructing a student academic performance matrix ,in, Indicates the first Period, No. Standardized average academic performance of students taught by each teacher This indicates the final grade. Indicates basic entrance examination scores. This represents the total number of students;

[0015] For the first Period, No. Teacher, No. All raw score data for each indicator Calculate its mean and standard deviation If the score of a questionnaire meets the requirements If a value is found to be an outlier, it will be removed, thus completing the data cleaning process.

[0016] The cleaned data was integrated into a related dataset. , This represents the original score matrix of teaching evaluations after cleaning. This represents the cleaned student academic performance matrix.

[0017] Further technical solution: In step S2:

[0018] For the Period, No. The teacher, obtain all of them The homogeneity coefficient of the scores of each evaluation indicator is calculated using the following formula:

[0019]

[0020] in, Indicates the first Period, No. The homogeneity coefficient of the evaluation data of individual teachers. Indicates the first The teacher in the The standard deviation of all indicator scores during the period Indicates the first The teacher in the The arithmetic mean of the scores of all indicators in the period;

[0021] Further technical solution: In step S2:

[0022] A homogenization correction factor is generated based on the homogenization coefficient, a preset homogenization threshold, and the total number of valid questionnaires. Specifically:

[0023]

[0024] in, Indicates the first Period, No. Homogeneity correction factor for teachers Represents the homogenization threshold, the first Period, No. The total number of valid questionnaires for each teacher is calculated by summing the number of valid questionnaires for all evaluation indicators of that teacher.

[0025] Further technical solution: In step S2:

[0026] Based on the original evaluation weights and original scores, and combined with a homogenization correction factor, the corrected individual evaluation scores are obtained:

[0027]

[0028] in, Indicates the first Period, No. The preliminary composite scores of the teachers after homogenization correction Indicates the first The preset weights of each evaluation indicator and , Indicates the first Period, No. Teacher, No. The original average score of each indicator Indicates the first Period, No. Homogeneity correction factor for teachers.

[0029] Further technical solution: In step S3:

[0030] Utilizing history Period data, for the first period Each evaluation indicator was used to obtain its score and the correlation coefficient between the score and academic performance.

[0031]

[0032] in, Indicates the first The correlation coefficient between the scores of each evaluation indicator and academic performance scores - grades. Indicates the first Period, No. Teacher, No. The original average score of each indicator Indicates the first Period, No. Standardized average academic performance of students taught by each teacher Indicates the first Historical average scores of each indicator This represents the historical average academic performance.

[0033] Further technical solution: In step S3:

[0034] The score-performance correlation coefficient is compared with a preset correlation coefficient threshold, and the adjusted unnormalized weights are obtained by combining the internal consistency reliability coefficient:

[0035]

[0036] in, Indicates the first The adjusted unnormalized weights of each evaluation indicator Indicates the first The preset weights of each evaluation indicator, This represents the correlation coefficient threshold. Indicates the first Internal consistency reliability coefficient of each evaluation indicator;

[0037] Among them, the The internal consistency reliability coefficient of each evaluation indicator is expressed as:

[0038]

[0039] in, Indicates the first The internal consistency reliability coefficient of the i-th evaluation index represents the i-th... The total number of scoring items under each evaluation indicator (usually the same for each period). This indicates the number of all historical cycles, all teachers, and all questionnaires. The variance is calculated by combining the scores of each rating item into a single sample. This indicates that all historical cycles, all teachers, and all questionnaires will be included in the first... The variance of the total scores under each evaluation indicator is calculated after merging them into a single sample.

[0040] Further technical solution: In step S3:

[0041] The unnormalized weights of all evaluation indicators are normalized to obtain the corrected weights:

[0042]

[0043] in, Indicates the first Adjusted weights for each evaluation indicator This is the sum of the unnormalized weights of all indicators.

[0044] Further technical solution: The specific steps of S4 are as follows:

[0045] Based on the latest evaluation cycle The original evaluation score, the first The final weights of the evaluation indicators after adjustment, along with the homogenization adjustment factor for the latest period, are used to obtain the [number]th evaluation indicator. The final overall evaluation score of each teacher:

[0046]

[0047] in, Indicates the first The final comprehensive evaluation score of each teacher, Indicates the latest evaluation period , No. Teacher, No. The original average score of each indicator Indicates the first Adjusted weights for each evaluation indicator Indicates the first Homogenization correction factor for each teacher in the latest evaluation cycle.

[0048] A further technical solution: The corrected weights calculated in step S3 are used to update the weight system for the next evaluation cycle, forming a dynamic iterative optimization closed loop for the evaluation index weights.

[0049] This invention provides a comprehensive evaluation method for teaching quality based on the reliability of teaching evaluation data, which has the following advantages compared with the prior art:

[0050] 1. This invention introduces a homogenization coefficient and the number of valid questionnaires to construct a homogenization correction factor, which effectively identifies and corrects data with poor scoring consistency, thereby improving the stability and credibility of individual evaluation scores.

[0051] 2. This invention combines historical evaluation data with student academic performance to calculate the score-performance correlation coefficient and integrates the internal consistency reliability coefficient to dynamically adjust the weight of each evaluation indicator, making the weight allocation more scientific and adaptable, and avoiding the bias caused by subjective weighting.

[0052] 3. This invention uses an outlier cleaning method to remove abnormal scores, ensuring the quality of basic data and providing reliable support for subsequent analysis;

[0053] 4. This invention forms a data-driven evaluation closed-loop mechanism. By iteratively optimizing the weight system through multiple periods of historical data, the evaluation method has the ability to continuously improve, thereby enhancing the accuracy and practicality of the comprehensive evaluation. Attached Figure Description

[0054] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0056] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0057] Please see Figure 1 The present invention provides a method for comprehensive evaluation of teaching quality based on the reliability of teaching evaluation data, comprising:

[0058] S1. Collect raw teaching evaluation scores and corresponding student academic performance data from multiple historical periods, clean up invalid data, and construct a related dataset containing teachers, evaluation indicators, and academic performance.

[0059] S2. Based on the associated dataset, obtain the homogenization coefficient of each teacher's evaluation indicator scores and the total number of valid questionnaires in each evaluation period. Based on the homogenization coefficient, the preset homogenization threshold and the total number of valid questionnaires, generate a homogenization correction factor. Use the homogenization correction factor to correct the teacher's original comprehensive score and output the homogenized individual score.

[0060] S3. Based on historical data from multiple periods in the associated dataset, obtain the score-grade correlation coefficient between the scores of each evaluation indicator and academic performance. Then, dynamically adjust the original weights of each evaluation indicator based on the score-grade correlation coefficient, the preset correlation coefficient threshold, and the internal consistency reliability coefficient of the evaluation indicator, and output the corrected indicator weights.

[0061] S4. Based on the original evaluation score of the latest evaluation cycle, the homogenization correction factor, and the corrected indicator weights, obtain and output the final comprehensive evaluation score of teaching quality for each teacher.

[0062] The following example will provide a more detailed explanation of the above technical solution:

[0063] Suppose an educational institution needs to conduct periodic evaluations of its teachers' teaching quality. The institution faces the following problems: traditional evaluation methods are easily influenced by students' arbitrary or extreme ratings, leading to distorted evaluation results; at the same time, the fixed weights of evaluation indicators fail to reflect the actual differences in the contributions of different indicators to students' academic performance; and furthermore, the evaluation results are not strongly correlated with students' actual academic performance, making it difficult to objectively reflect teaching quality.

[0064] To address the aforementioned issues, the method provided in this embodiment is applied to the educational institution.

[0065] First, in step S1, the system automatically collects raw teaching evaluation scores and corresponding student academic performance data from the institution's academic management system and student information system over the past five years (multiple historical periods). For example, for a course taught by Teacher A in the spring semester of 2020, the system obtains all students' evaluation questionnaire scores, as well as their final exam scores and initial entrance scores. Subsequently, the system cleans this raw data, identifying and removing questionnaire data with clearly abnormal scoring behavior (such as all indicators receiving the highest or lowest score, or excessively large scoring variance), and data entry errors in the academic performance data. After cleaning, this data is integrated into a linked dataset containing information such as teacher ID, evaluation period, evaluation indicator scores, and student academic performance. This dataset provides a reliable foundation for subsequent analysis, avoiding evaluation bias caused by problems with the quality of the raw data.

[0066] Secondly, in step S2, based on the associated dataset, the system performs homogenization analysis on the evaluation data of each teacher. For example, for teacher B's evaluation data in the fall semester of 2021, the system calculates the homogenization coefficient of all evaluation indicator scores, which reflects the consistency of students' ratings of teacher B across different evaluation dimensions. Simultaneously, the system counts the total number of valid questionnaires for that teacher during that period. Assuming teacher B has a high homogenization coefficient and a sufficient number of valid questionnaires, the generated homogenization correction factor is close to 1. Conversely, if teacher C's evaluation data in the spring semester of 2022 has a low homogenization coefficient (e.g., students rated some indicators highly and others poorly, showing inconsistency), or the total number of valid questionnaires is insufficient, the system generates a homogenization correction factor less than 1. This correction factor is then applied to the teacher's original composite score, correcting downwards for evaluation results with poor rating consistency or insufficient sample size, thereby effectively reducing the impact of extreme or arbitrary ratings on the final evaluation results and improving the fairness and credibility of the evaluation.

[0067] Next, in step S3, the system dynamically adjusts the weights of the evaluation indicators using historical multi-period data. For example, the system analyzes the correlation between the scores of each evaluation indicator (such as "organization of teaching content," "classroom interaction," and "homework grading") and student academic performance over the past five years. If the score of the "organization of teaching content" indicator shows a significant positive correlation with student academic performance, indicating that the indicator has a strong predictive power for teaching effectiveness, the system will dynamically increase the weight of the "organization of teaching content" indicator based on the score-performance correlation coefficient, a preset correlation coefficient threshold, and the internal consistency reliability coefficient of the indicator. Conversely, if an indicator has no significant correlation with academic performance, or its internal consistency reliability is low, its weight will be reduced accordingly or set to zero. In this way, the weights of the evaluation indicators are no longer fixed but can be optimized according to their actual correlation with teaching effectiveness, making the evaluation results more reflective of the actual contribution of the indicators.

[0068] Finally, in step S4, the system calculates the final comprehensive evaluation score for each teacher's teaching quality based on the original evaluation scores for the latest evaluation period (e.g., the fall semester of 2023), the homogenization correction factor generated for each teacher in step S2, and the dynamically adjusted indicator weights in step S3. For example, for teacher D, the system weights and sums the original average scores of each indicator in the fall semester of 2023 with the corrected indicator weights, and then multiplies it by the homogenization correction factor for that period to obtain a comprehensive final evaluation score that has been corrected for credibility. This score not only considers the students' original evaluations but also corrects for the credibility of the scores through the homogenization correction factor and reflects the actual contribution of each indicator to academic performance through the dynamically adjusted indicator weights, making the final evaluation results more objective, scientific, and effective.

[0069] Existing technologies for evaluating teaching quality often neglect the internal consistency of evaluation data, failing to effectively identify and correct for homogeneity issues in ratings from different teachers or across different periods. This makes evaluation results susceptible to the influence of extreme or arbitrary ratings. However, this embodiment introduces a homogeneity coefficient and a homogeneity correction factor in step S2 to correct the teacher's original comprehensive score. For example, in the above example, when teacher C's evaluation data has a low homogeneity coefficient or insufficient total number of valid questionnaires, the system generates a homogeneity correction factor less than 1 to lower the original score. This mechanism effectively avoids evaluation bias caused by inconsistent ratings or insufficient sample size, significantly improving the fairness and credibility of the evaluation results—something existing fixed weighted summation methods lack.

[0070] Furthermore, existing technologies typically lack a dynamic adjustment mechanism for the weighting of evaluation indicators, failing to optimize weights based on historical data and teaching effectiveness, thus making it difficult to accurately reflect the actual contribution of the indicators. In contrast, this embodiment, in step S3, obtains the score-performance correlation coefficient between the scores of each evaluation indicator and student academic performance based on historical multi-period data, and dynamically adjusts the original weights of each evaluation indicator in conjunction with the internal consistency reliability coefficient. For example, in this example, the "organization of teaching content" indicator is given a higher weight because it shows a significant positive correlation with student academic performance. This dynamic adjustment mechanism allows the weights of evaluation indicators to be optimized according to their actual correlation with teaching effectiveness, ensuring that the evaluation results better reflect the true contribution of each indicator and overcoming the rigidity of weighting settings in existing technologies.

[0071] Furthermore, existing technologies generally lack systematic analysis of the correlation between evaluation data and academic performance, leading to discrepancies between evaluation results and the true level of teaching quality, thus affecting the objectivity and effectiveness of the evaluation. This embodiment integrates the correlation analysis between evaluation indicators and academic performance into weight adjustment and final score calculation through steps S3 and S4. For example, in this example, the final comprehensive teaching quality evaluation score not only considers the original evaluation but also reflects the actual contribution of each indicator to academic performance through dynamically adjusted indicator weights, and corrects for the credibility issue of the score through a homogenization correction factor. Therefore, the method provided in this embodiment can output a comprehensive evaluation score that has undergone credibility correction and is closely related to teaching effectiveness, making the evaluation results more objective, scientific, and effective, thereby overcoming the shortcomings of existing technologies in terms of evaluation objectivity and effectiveness.

[0072] Preferably, the specific steps of S1 are as follows:

[0073] Constructing the original score matrix for teaching evaluation ,in, Indicates the first Period, No. Teacher, No. The original average score of each indicator For the first Valid questionnaire scores This indicates the total number of valid questionnaires;

[0074] Constructing a student academic performance matrix ,in, Indicates the first Period, No. Standardized average academic performance of students taught by each teacher This indicates the final grade. Indicates basic entrance examination scores. This represents the total number of students;

[0075] For the first Period, No. Teacher, No. All raw score data for each indicator Calculate its mean and standard deviation If the score of a questionnaire meets the requirements If a value is found to be an outlier, it will be removed, thus completing the data cleaning process.

[0076] The cleaned data was integrated into a related dataset. , This represents the original score matrix of teaching evaluations after cleaning. This represents the cleaned student academic performance matrix.

[0077] The purpose of constructing the original score matrix for teaching evaluation is to represent the original score data of teaching evaluation from multiple periods, teachers, and indicators in a structured manner to facilitate subsequent data processing and analysis. This can be achieved by storing the data in a specific table of a relational database, or by using data structures such as multidimensional arrays or data frames in the data processing program. Indicates the first Period, No. Teacher, No. The original average score of each indicator is calculated in a way that ensures that the score of each indicator can comprehensively reflect the opinions of all valid questionnaires. It can be calculated automatically through the data collection system's backend or by using data analysis tools (such as Python's Pandas library). For the first A valid questionnaire score refers to the score provided by a questionnaire that has passed the initial screening and meets the validity criteria. Its validity can be ensured by setting mandatory fields and logical verification rules during the questionnaire design, or by conducting a preliminary screening manually or procedurally after data collection. The total number of valid questionnaires is the basis for calculating the average score. It can be obtained by counting the questionnaires that meet the criteria during the data collection and screening process, or by counting the number of records in the database through a query statement.

[0078] The purpose of constructing a student academic performance matrix is ​​to provide a structured representation of academic performance data from students taught by multiple teachers across different periods, facilitating correlation analysis with teaching evaluation score matrices. This can be achieved by storing students' final exam scores and initial entrance scores in a database and then constructing the matrix programmatically, or by directly creating the corresponding data structure within a data analysis tool. Indicates the first Period, No. The standardized average academic performance of students taught by a teacher is calculated by subtracting the initial entrance score from the final exam score to measure students' academic progress. The average of the progress of all students taught by the teacher helps to eliminate the impact of differences in students' initial entrance scores on the evaluation of teaching effectiveness. Batch calculation can be performed using programming scripts (such as SQL queries or Python scripts). Final grades are an important indicator of students' learning outcomes and can be obtained directly from the academic management system or grade database. These represent the basic academic performance for admission and are a key reference for assessing the contribution of teaching to students' academic progress. They can be obtained from the admissions system, entrance examination score database, or student records. The total number of students represents the basis for calculating the standardized average academic performance, which can be obtained from the academic management system or course selection records.

[0079] For the first Period, No. Teacher, No. All original score data for each indicator Calculate its mean and standard deviation These statistics are key to identifying outliers and can be used to perform descriptive statistical analysis on the data using statistical analysis software (such as SPSS, R, and Python's SciPy library). If the scores of a questionnaire meet... If any values ​​are found to be outliers, they are removed, completing the data cleaning process. This outlier detection and removal method based on the "3 Sigma" principle can effectively remove extreme and unreasonable scores, improving data quality and reliability. Automated detection and removal can be achieved through programming scripts. The cleaned data is then integrated into a related dataset. , This represents the original score matrix of teaching evaluations after cleaning. This represents the cleaned student academic performance matrix. This step merges the teaching evaluation score data (after outlier removal) and student academic performance data into a unified associated dataset, ensuring data integrity and consistency. This can be achieved by creating an associated table in a database or by merging the two cleaned data frames in a data analysis environment.

[0080] This application's solution ensures the accuracy and reliability of subsequent comprehensive evaluations of teaching quality by structuring and meticulously cleaning the original teaching evaluation data and student academic performance data. First, by constructing an original teaching evaluation score matrix and a student academic performance matrix, the scattered original data is given a unified, structured, and standardized representation. The original teaching evaluation score matrix obtains the original average score for each teacher on each evaluation indicator by averaging the scores of each valid questionnaire, providing foundational data for subsequent homogenization correction and weight adjustment. Simultaneously, the student academic performance matrix achieves a standardized measurement of student academic progress by calculating and averaging the difference between students' final exam scores and their initial entrance scores, effectively eliminating the interference of individual student differences in teaching effectiveness evaluation, and enabling the academic performance data to more objectively reflect teachers' teaching effectiveness. Building on this, this application introduces a data cleaning mechanism based on statistical principles: for all original score data of each teacher, each evaluation period, and each evaluation indicator, the mean and standard deviation are calculated, and the "3 Sigma" principle is used to detect and remove outliers from the score data. This cleaning method effectively identifies and removes extreme and unreasonable evaluation data, such as outliers caused by malicious scoring or misoperation, thereby significantly improving the quality and credibility of the original teaching evaluation score data. The cleaned original teaching evaluation score matrix and student academic performance matrix are integrated into a linked dataset. This dataset not only contains high-quality teaching evaluation scores and academic performance data, but also, through its structured form, provides a unified, accurate, and reliable data foundation for subsequent steps such as calculating the homogenization coefficient, generating correction factors, and dynamically adjusting indicator weights, thus ensuring the scientific validity and effectiveness of the entire comprehensive teaching quality evaluation method.

[0081] The following is a concrete example. In practical applications, the school's online teaching evaluation system can be used to collect student evaluation data on teachers, and students' final grades and initial grades upon enrollment can be obtained from the academic affairs management system. For example, when constructing the raw teaching evaluation score matrix, the system can extract all student rating records for a teacher in a given semester for the "teaching attitude" indicator from the database. Assuming that the teacher has 80 valid questionnaire ratings for this indicator, the system will sum these 80 ratings and divide by 80 to obtain the raw average score for this indicator. For constructing the student academic performance matrix, the system can obtain the final exam scores and standardized test scores of the students in the teacher's classes from the student grade database. For example, if the teacher's class has 45 students, the system will calculate the difference between each student's final grade and their initial grade, then sum these 45 differences and divide by 45 to obtain the standardized average academic performance of the students taught by the teacher. The data cleaning process can be performed procedurally. For example, once the system obtains all the raw score data for a teacher under a certain indicator in a certain semester, it can use data processing software to calculate the mean and standard deviation of these scores. Then, an automated script checks each raw score. If the absolute difference between a score and the calculated mean is greater than three times the standard deviation, the score is marked as an outlier and removed from the dataset. For example, if the mean of an indicator is 88 points and the standard deviation is 4 points, any score below 76 (88 - 3 × 4) or above 100 (88 + 3 × 4) will be discarded. After cleaning all indicators and teacher scores, the cleaned raw teaching evaluation scores and student academic performance data are stored in a unified data warehouse, forming a related dataset. For example, it can be stored as multiple related tables in a relational database for efficient processing in subsequent steps.

[0082] By employing the aforementioned structured construction and refined cleaning methods for the raw data, this application effectively addresses the problem of distorted evaluation results caused by low-quality raw data. Specifically, by constructing a raw teaching evaluation score matrix and a student academic performance matrix, unified management and standardized representation of multi-source heterogeneous data are achieved, providing a clear and well-organized data foundation for subsequent complex calculations. More importantly, the introduction of an outlier removal mechanism based on mean and standard deviation accurately identifies and removes extreme abnormal scores, avoiding interference from malicious evaluations or random errors in the teaching quality evaluation results, and significantly improving the authenticity and reliability of the teaching evaluation data. This data preprocessing method ensures the accuracy of subsequent homogenization correction factors and indicator weight adjustments, thereby enabling the final comprehensive teaching quality evaluation score to more objectively and fairly reflect the teacher's actual teaching level and effectiveness.

[0083] Preferably, the specific steps of S2 are as follows:

[0084] For the Period, No. The teacher, obtain all of them The homogeneity coefficient of the scores of each evaluation indicator is calculated using the following formula:

[0085]

[0086] in, Indicates the first Period, No. The homogeneity coefficient of the evaluation data of individual teachers. Indicates the first The teacher in the The standard deviation of all indicator scores during the period Indicates the first The teacher in the The arithmetic mean of the scores of all indicators in the period;

[0087] A homogenization threshold is set, and a homogenization correction factor is generated by comparing the homogenization coefficient with the homogenization threshold. Specifically:

[0088]

[0089] in, Indicates the first Period, No. Homogeneity correction factor for teachers Represents the homogenization threshold, the first Period, No. The total number of valid questionnaires for each teacher is calculated by summing the number of valid questionnaires for all evaluation indicators of that teacher.

[0090] Based on the original evaluation weights and original scores, and combined with a homogenization correction factor, the corrected individual evaluation scores are obtained:

[0091]

[0092] in, Indicates the first Period, No. The preliminary composite scores of the teachers after homogenization correction Indicates the first The preset weights of each evaluation indicator and , Indicates the first Period, No. Teacher, No. The original average score of each indicator Indicates the first Period, No. Homogeneity correction factor for teachers.

[0093] This step aims to quantify the dispersion of teachers' scores across all evaluation indicators within a given evaluation period. Homogeneity coefficient. The calculation method is standard deviation. Divide by the arithmetic mean Among them, standard deviation Reflects the first The teacher in the The volatility of all indicator scores over the period, while the arithmetic mean This represents the overall score level. This ratio effectively measures the consistency or central tendency of the evaluation data. For example, when teachers' scores on various indicators are very close, the homogeneity coefficient will be small, indicating that the evaluation data tends to be homogeneous; conversely, if the scores on various indicators differ significantly, the homogeneity coefficient will be large, indicating that the evaluation data has good discriminative power. Generating a homogeneity correction factor... The steps are used to determine the homogenization coefficient. Preset homogenization threshold and the total number of valid questionnaires To dynamically adjust the correction factor. When the homogeneity coefficient Less than the preset homogenization threshold When this occurs, it indicates that the evaluation data has high homogeneity, and the correction factor is needed. The correction factor is calculated based on the degree of homogeneity and the number of valid questionnaires to reduce the influence of the raw score. For example, the correction factor can be designed to be directly proportional to the homogeneity coefficient and inversely proportional to the logarithm of the total number of valid questionnaires, thus providing a greater correction when homogeneity is high and the number of questionnaires is large. If the homogeneity coefficient... Not less than the homogenization threshold If so, the evaluation data is considered to have sufficient discriminative power, and the correction factor is applied. Set to 1, meaning no correction is performed. Homogenization threshold. This is a preset constant that can be set based on experience in actual teaching evaluation and the characteristics of data distribution to balance the sensitivity and stability of the evaluation. The corrected individual evaluation score is then obtained. The steps to change the original evaluation weights Compared with the original score The weighted sum, and the generated homogenization correction factor By combining these factors, a revised individual evaluation score can be obtained. Original evaluation weights Indicates the first The sum of the importance of each evaluation indicator in the overall evaluation is 1. Raw Score It is the first Period, No. Teacher, No. The original average score of each indicator. This is achieved by multiplying the original weighted score by a homogenization correction factor. This method can effectively adjust the scores of teachers whose evaluation data is too homogeneous, making them more reflective of their true teaching level. For example, when the homogeneity correction factor is less than 1, the corrected score will be lower than the original weighted score, thus penalizing teachers whose evaluation data lacks differentiation; when the correction factor is 1, the original weighted score remains unchanged.

[0094] This application's solution addresses the potential homogeneity issue in teaching evaluation data by introducing a homogenization coefficient and a homogenization correction factor to correct teachers' original comprehensive scores. Specifically, firstly, for each teacher's scores across all evaluation indicators within a specific evaluation period, a homogenization coefficient is calculated. This coefficient reflects the discriminative power of the evaluation data by measuring the dispersion of scores across different indicators. A low homogenization coefficient indicates that student evaluations tend to concentrate on a specific score range, lacking detailed differentiation across different teaching dimensions, thus indicating homogenization in the evaluation data. Based on this, the system generates a homogenization correction factor using the calculated homogenization coefficient, a preset homogenization threshold, and the total number of valid questionnaires. This correction factor aims to appropriately adjust evaluation results with high homogenization levels. When the homogenization coefficient is below the preset threshold, the correction factor will be less than 1, and its value will further decrease as the degree of homogenization increases and the number of valid questionnaires increases, resulting in a more significant downward adjustment of the original score. Conversely, if the homogenization coefficient reaches or exceeds the threshold, the evaluation data is considered to have sufficient discriminative power, the correction factor remains at 1, and no correction is made to the original score. Finally, the weighted sum of the teacher's original evaluation weights and original scores is multiplied by the homogenization correction factor to obtain the corrected individual evaluation score. This series of steps ensures that when calculating the teacher's preliminary comprehensive score, it can effectively identify and reduce inflated or inaccurate scores caused by homogenization of evaluation data, making the final evaluation result more objective and fair, thereby improving the reliability of the comprehensive evaluation of teaching quality.

[0095] The following is a specific example to illustrate this. Suppose that during a certain evaluation period... In China, targeting teachers Its teaching evaluation includes =5 indicators. First, the system will collect all student rating data for the teacher on these 5 indicators and calculate the average score for each indicator. Next, based on the average scores of these five indicators, their standard deviations were calculated. and arithmetic mean For example, if the average scores of these five indicators are 90, 92, 91, 89, and 93 respectively, their standard deviation and average can be calculated, and thus the homogeneity coefficient can be obtained. Assuming the calculated The value is 0.03. Meanwhile, the preset homogenization threshold... It can be set to 0.05, the total number of valid questionnaires. There are 100 copies. Because... (0.03) Less than (0.05), the system will use the formula Calculate the homogenization correction factor =0.13. Assume the original evaluation weights are... If the values ​​are 0.2, 0.2, 0.2, 0.2, and 0.2 respectively, then the original weighted composite score is... = 91. Finally, the revised individual rating score. It will be the original weighted composite score of 91 multiplied by the homogenization correction factor of 0.13, that is... =11.83. This correction process significantly reduced the potentially inflated original scores due to the homogenization of evaluation data, making the evaluation results more valuable for reference.

[0096] Through the above technical solution, this application can effectively identify and address the homogenization problem in teaching evaluation data. By calculating the homogenization coefficient, the concentration of student evaluations can be quantified, avoiding distortion of evaluation results due to convergence of student evaluations. Furthermore, a homogenization correction factor is generated based on the homogenization coefficient, a preset threshold, and the total number of valid questionnaires. This factor can reasonably correct teacher scores with high homogenization of evaluation data, thereby reducing the impact of evaluations lacking differentiation on the final comprehensive score. This allows the corrected individual evaluation scores to more accurately reflect teachers' teaching levels, improving the accuracy and fairness of the comprehensive evaluation of teaching quality, avoiding evaluation bias caused by insufficient reliability of evaluation data, and providing a more reliable basis for teaching management and teacher development.

[0097] Preferably, the specific steps of S3 include:

[0098] Utilizing history Period data, for the first period Each evaluation indicator was used to obtain its score and the correlation coefficient between the score and academic performance.

[0099]

[0100] in, Indicates the first The correlation coefficient between the scores of each evaluation indicator and academic performance scores - grades. Indicates the first Period, No. Teacher, No. The original average score of each indicator Indicates the first Period, No. Standardized average academic performance of students taught by each teacher Indicates the first Historical average scores of each indicator This represents the historical average academic performance.

[0101] The score-performance correlation coefficient is compared with a preset correlation coefficient threshold, and the adjusted unnormalized weights are obtained by combining the internal consistency reliability coefficient:

[0102]

[0103] in, Indicates the first The adjusted unnormalized weights of each evaluation indicator Indicates the first The preset weights of each evaluation indicator, This represents the threshold for the correlation coefficient. Indicates the first Internal consistency reliability coefficient of each evaluation indicator;

[0104] Among them, the The internal consistency reliability coefficient of each evaluation indicator is expressed as:

[0105]

[0106] in, Indicates the first The internal consistency reliability coefficient of the i-th evaluation index represents the i-th... The total number of scoring items under each evaluation indicator (usually the same for each period). This indicates the number of all historical cycles, all teachers, and all questionnaires. The variance is calculated by combining the scores of each rating item into a single sample. This indicates that all historical cycles, all teachers, and all questionnaires will be included in the first... The variance of the total scores under each evaluation indicator is calculated after merging them into a single sample.

[0107] The unnormalized weights of all evaluation indicators are normalized to obtain the corrected weights:

[0108]

[0109] in, Indicates the first Adjusted weights for each evaluation indicator This is the sum of the unnormalized weights of all indicators.

[0110] The purpose of obtaining the score-performance correlation coefficient is to quantify the statistical strength of the association between the scores of each evaluation indicator and students' academic performance. This can be achieved through various statistical methods. For example, the Pearson correlation coefficient can be used to measure the linear relationship between two variables, or the Spearman rank correlation coefficient can be used to assess non-linear or rank relationships. By calculating using historical data from multiple periods, it is possible to reveal which teaching evaluation indicators have a stronger positive correlation with students' academic performance.

[0111] The comparison of the score-performance correlation coefficient with a preset correlation coefficient threshold, combined with the internal consistency reliability coefficient to obtain the adjusted unnormalized weight, serves to dynamically adjust the importance of the evaluation indicator in the comprehensive evaluation based on its actual effectiveness and reliability. Specifically, a correlation coefficient threshold can be set, such as 0.1 or 0.2. If the score-performance correlation coefficient of an evaluation indicator is lower than this threshold, the indicator is considered to have an insignificant association with academic performance, and its weight can be set to zero, thus weakening or eliminating it from the evaluation. For indicators whose correlation coefficient reaches or exceeds the threshold, their preset weight will be further adjusted based on their internal consistency reliability coefficient. The internal consistency reliability coefficient reflects the degree of consistency among the various scoring items within the evaluation indicator; for example, it can be obtained through Cronbach's alpha. The coefficient (Cronbach's Alpha) is used to calculate the index. The higher the coefficient, the more stable and reliable the measurement of the index.

[0112] The internal consistency reliability coefficient measures the consistency or reliability among the rating items within an evaluation index. This coefficient is typically calculated based on analysis of variance; for example, a Cronbach's alpha algorithm can be used. The coefficient is calculated by comparing the variance of each rating item with the total variance. A high internal consistency reliability coefficient indicates that the components of the evaluation index exhibit high consistency in measuring the same concept, thereby enhancing the measurement validity of the index.

[0113] The normalization process for all unnormalized weights of the evaluation indicators yields corrected weights. This corrected weighting transforms the adjusted weights of each evaluation indicator into proportions that sum to 1, ensuring that the weight allocation for all indicators is reasonable and that the sum is 1 in the final comprehensive score calculation. This can be achieved by dividing each unnormalized weight by the sum of all unnormalized weights, thus obtaining corrected weights between 0 and 1.

[0114] In the methods described above, although the teaching evaluation data has been cleaned and homogenized, the weights of the evaluation indicators may still use preset fixed values. This could lead to the evaluation results failing to fully reflect the actual correlation between each indicator and student academic performance. To address this, this application further proposes a method for dynamically adjusting the weights of evaluation indicators. First, it utilizes historical data... Based on the data from the previous period, the system will calculate the correlation coefficient between the score of each evaluation indicator and the student's academic performance score - grade. This step aims to objectively reveal, through big data analysis, which teaching behaviors or evaluation dimensions have significant statistical correlations with student learning outcomes. For example, if the score for the "classroom interaction" indicator shows a high positive correlation with student grades, it indicates that this indicator is an important dimension for measuring teaching quality. Subsequently, the score-grade correlation coefficient will be calculated. Correlation coefficient threshold with preset A comparison is then performed. This comparison mechanism effectively filters out evaluation indicators that have a low correlation with student academic performance, and adjusts their unnormalized weights accordingly. Setting it to zero avoids irrelevant or weakly correlated indicators from interfering with the final evaluation results. Meanwhile, for those correlation coefficients that reach or exceed the threshold... The indicators, with their preset weights This will be further combined with its internal consistency reliability coefficient. Adjustments were made. Internal consistency reliability coefficient. This is used to assess the reliability and stability of the evaluation indicators themselves, ensuring that the indicators used are effective and consistent measurement tools. In this way, not only the external correlation between the indicators and academic performance is considered, but also the intrinsic quality of the indicators themselves. Finally, the unnormalized weights are adjusted using both correlation coefficients and reliability coefficients. The normalization process will be performed to obtain the corrected weights. This series of steps ensures that the weights of the evaluation indicators are no longer static or subjectively predetermined, but are dynamically adjusted based on the actual correlation between the indicators and students' academic performance in historical data, as well as the measurement reliability of the indicators themselves. This data-driven weight adjustment mechanism enables a more accurate reflection of the true contribution of each teaching dimension to students' learning outcomes when calculating the teacher's final comprehensive teaching quality evaluation score.

[0115] As a specific implementation method, suppose an educational institution needs to adjust the weights of three evaluation indicators: "teaching methods," "course content," and "student participation." First, the system will utilize teaching evaluation data from the past five years (historical period T) and corresponding student academic performance data. For the "teaching methods" indicator, it will calculate the correlation coefficient between its score and the standardized average academic performance score (SAP). Additionally, if the "Teaching Methods" category includes multiple rating items such as "heuristic teaching" and "case-based teaching," the internal consistency reliability coefficient among these rating items will be calculated. Assuming a preset correlation coefficient threshold. It is 0.2. If the calculation yields "teaching methods"... It is 0.6, and Its preset weight is 0.85. It is 0.4. Therefore, according to the formula... The adjusted unnormalized weights of "teaching methods" can be calculated. If the correlation coefficient between the score and grade of the "Course Content" indicator... The correlation coefficient is only 0.15, which is lower than the preset threshold. Then its adjusted unnormalized weights This will be set directly to 0. This means that, based on the current data, this indicator is considered not significantly related to student academic performance, and its influence in the overall evaluation is eliminated. For the "student participation" indicator, assuming its... It is 0.4. The preset weight is 0.75. The value is 0.3. Similarly, its adjusted unnormalized weight can be calculated. After calculating the unnormalized weights of all indicators... , , The system will then normalize these weights, i.e., calculate... This leads to the final corrected weights used for comprehensive evaluation of teaching quality. , , For example, if =0.2, =0, If the sum is 0.1, then the total is 0.3, and the adjusted weights will be respectively... =0.67, =0, =0.33. In this way, the weights of the evaluation indicators can be dynamically adjusted based on their actual correlation with students' academic performance and their own reliability, making the evaluation results more convincing.

[0116] The above method, based on data cleaning and homogenization correction, further addresses the limitations of traditional teaching quality evaluation in terms of indicator weighting. By utilizing historical data from multiple periods, it calculates the score-grade correlation coefficient between the scores of each evaluation indicator and students' academic performance. This application can objectively quantify the actual impact of each teaching dimension on student learning outcomes. Furthermore, it incorporates preset correlation coefficient thresholds. Internal consistency reliability coefficient of evaluation indicators To dynamically adjust the original weights of each evaluation indicator This data-driven weighting adjustment mechanism assigns higher weights to indicators that are highly correlated with student academic performance and have high self-measurement reliability, while reducing or even eliminating indicators with low correlation or insufficient reliability. This overcomes the subjectivity and inaccuracy that may arise from traditional fixed weights, enabling the final comprehensive teaching quality evaluation score to more accurately and objectively reflect teachers' teaching effectiveness. This significantly improves the scientific validity and effectiveness of the evaluation results, providing a more reliable basis for teaching improvement.

[0117] Preferably, the specific steps of S4 are as follows:

[0118] Based on the latest evaluation cycle The original evaluation score, the first The final weights of the evaluation indicators after adjustment, along with the homogenization adjustment factor for the latest period, are used to obtain the [number]th evaluation indicator. The final overall evaluation score of each teacher:

[0119]

[0120] in, Indicates the first The final comprehensive evaluation score of each teacher, Indicates the latest evaluation period , No. Teacher, No. The original average score of each indicator Indicates the first Adjusted weights for each evaluation indicator Indicates the first Homogenization correction factor for each teacher in the latest evaluation cycle.

[0121] Among them, the latest evaluation cycle Original evaluation score This refers to the unadjusted average score of each teacher across all evaluation indicators within the current or most recent evaluation period. This score directly reflects the students' or evaluators' direct perception and evaluation of the teacher on specific indicators, and is the foundational data for teaching quality evaluation. It can be obtained through automatic collection and aggregation via an online teaching evaluation system, or through manual input and calculation. The final weights of each evaluation indicator after correction This refers to the weights of each evaluation indicator after dynamic adjustment through historical data analysis (e.g., correlation analysis with student academic performance, internal consistency reliability calculation). These weights reflect the relative importance of different evaluation indicators in teaching quality evaluation, and this importance is optimized based on data reliability. Its purpose is to ensure that the weights of the evaluation indicators can more objectively and scientifically reflect their contribution to teaching quality. These weights can be calculated using a preset weight adjustment algorithm (as described in step S3 of the above method) or dynamically adjusted using an expert system combined with historical data. The latest cycle's homogenization correction factor. This refers to a factor that corrects for the homogeneity (i.e., consistency or dispersion of evaluation scores) of evaluation data for each teacher within the latest evaluation period. This factor measures the reliability of the evaluation data, preventing distortion caused by overly concentrated or dispersed data. Its function is to correct teachers' original composite scores, making them more reflective of their true teaching level. This factor can be generated by calculating the ratio of the standard deviation to the mean of the evaluation indicator scores (as described in step S2 of the above method) and combining it with the total number of valid questionnaires, or by using other statistical methods (such as the coefficient of variation) to quantify the homogeneity of the evaluation data. The final comprehensive evaluation score of each teacher This refers to the final evaluation result obtained by comprehensively calculating the original evaluation scores, the revised indicator weights, and the homogenization correction factor. This score represents the overall teaching quality level of teachers within the latest evaluation period, aiming to provide a comprehensive, objective, and reliable evaluation result. It is calculated by weighted summation and multiplication by the correction factor (as shown in the formula above), or by using other multi-factor comprehensive evaluation models.

[0122] This application's approach integrates information from multiple sources to ensure the comprehensiveness and credibility of the final evaluation results, based on the latest evaluation cycle. Original evaluation score As the foundational data for evaluation, this directly reflects students' or evaluators' perceptions of teachers' teaching performance. Based on this, a revised final weighting is introduced. These weights are dynamically adjusted based on the correlation between evaluation indicators and student academic performance in historical data, as well as the internal consistency reliability coefficients of the indicators themselves. This weight adjustment mechanism allows the evaluation system to more scientifically identify indicators that truly contribute to teaching quality and assign them corresponding weights, thus avoiding the bias that may arise from fixed weights. Subsequently, the latest cycle's homogenization correction factor is used... The weighted scores are corrected using a correction factor that identifies and quantifies the homogeneity of the evaluation data itself, i.e., the consistency of the evaluation scores. This effectively suppresses evaluation distortion caused by overly concentrated or dispersed evaluation data, improving the reliability of the evaluation results. In this way, the method organically combines the original evaluation data, the reliability-verified indicator weights, and the homogeneity considerations of the evaluation data itself. Specifically, it first utilizes the corrected indicator weights... For the latest evaluation cycle The raw average score of each teacher on each indicator A weighted sum is performed to obtain a preliminary comprehensive score. This preliminary comprehensive score is then compared with the teacher's homogeneity correction factor in the latest evaluation period T. Multiply them to obtain the final comprehensive evaluation score. This calculation process ensures that the final score not only reflects the teacher's performance on each indicator, but also takes into account the actual utility of these indicators and the credibility of the evaluation data, thus solving the problem of inaccurate or unreliable evaluations that may occur when relying solely on raw evaluation data and fixed weights.

[0123] The following is a concrete example to illustrate this. Suppose a university's teaching evaluation system needs to conduct annual teaching quality evaluations of its teachers. In the latest evaluation cycle... The system first summarizes the feedback from student evaluation questionnaires to obtain each teacher's feedback. The raw average score on each evaluation indicator (e.g., teaching content, teaching methods, classroom management, teacher-student interaction, teaching effectiveness, etc.) Simultaneously, the system will obtain each evaluation indicator based on historical teaching data, academic performance data, and the calculation results of step S3 in the above method. Final weights after correction For example, the adjusted weight for the teaching content indicator might be 0.25, the adjusted weight for the teaching method indicator might be 0.20, and the adjusted weight for the teaching effectiveness indicator might be 0.30. Furthermore, the system will also determine the teacher's performance in the latest evaluation cycle based on the calculation results of step S2 in the above method. Homogenization correction factor This factor might be 0.98, indicating high homogeneity in the evaluation data. The system then substitutes this data into the formula for calculation. For example, if a teacher's original average score for the teaching content indicator is 4.6, for the teaching method indicator it's 4.3, and for the teaching effectiveness indicator it's 4.7, then their final comprehensive evaluation score... The calculation will be performed as follows: First, multiply the original average score of each indicator by its adjusted weight and sum them, i.e., (4.6×0.25)+(4.3×0.20)+(4.7×0.30)+... Then, multiply this weighted sum by a homogenization correction factor of 0.98 to obtain the teacher's final comprehensive evaluation score.

[0124] Through the above technical solution, this application can provide a more accurate and reliable comprehensive evaluation result of teaching quality. When relying solely on raw evaluation data and fixed weights for evaluation, there may be problems such as insufficient homogeneity of evaluation data (e.g., student evaluations being too concentrated or too dispersed) and unreasonable weighting of evaluation indicators (e.g., some indicators with low correlation to teaching effectiveness being given high weights), leading to the final evaluation result failing to truly reflect the teacher's teaching level. This application, by considering not only the raw evaluation score of the latest evaluation period when calculating the final comprehensive evaluation score, also introduces indicator weights verified and corrected by historical data, as well as correction factors to address the homogeneity of evaluation data. This comprehensive consideration ensures that the final evaluation score... This effectively avoids potential biases and uncertainties in the original evaluation data, ensuring the scientific rigor and objectivity of the evaluation results. Specifically, the corrected indicator weights... This ensures that indicators that truly impact teaching quality receive more reasonable attention, while homogenization correction factors... This effectively filters out interference caused by insufficient reliability of evaluation data, thereby significantly improving the accuracy and reliability of the comprehensive evaluation of teaching quality and providing more solid data support for teaching management and teacher development.

[0125] Preferably, the corrected weights calculated using historical multi-period data in step S3 are used to update the weight system for the next evaluation period, forming a dynamic iterative optimization closed loop of evaluation index weights.

[0126] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus.

[0127] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A comprehensive evaluation method for teaching quality based on the reliability of teaching evaluation data, characterized in that, include: S1. Collect raw teaching evaluation scores and corresponding student academic performance data from multiple historical periods, clean up invalid data, and construct a related dataset containing teachers, evaluation indicators, and academic performance. S2. Based on the associated dataset, obtain the homogenization coefficient of each teacher's evaluation indicator scores and the total number of valid questionnaires in each evaluation period. Based on the homogenization coefficient, the preset homogenization threshold and the total number of valid questionnaires, generate a homogenization correction factor. Use the homogenization correction factor to correct the teacher's original comprehensive score and output the homogenized individual score. S3. Based on historical data from multiple periods in the associated dataset, obtain the score-grade correlation coefficient between the scores of each evaluation indicator and academic performance. Then, dynamically adjust the original weights of each evaluation indicator based on the score-grade correlation coefficient, the preset correlation coefficient threshold, and the internal consistency reliability coefficient of the evaluation indicator, and output the corrected indicator weights. S4. Based on the original evaluation score of the latest evaluation cycle, the homogenization correction factor, and the corrected indicator weights, obtain and output the final comprehensive evaluation score of teaching quality for each teacher.

2. The comprehensive evaluation method for teaching quality based on the credibility of teaching evaluation data according to claim 1, characterized in that, The specific steps of S1 are as follows: Constructing the original score matrix for teaching evaluation ,in, Indicates the first Period, No. Teacher, No. The original average score of each indicator For the first Valid questionnaire scores This indicates the total number of valid questionnaires; Constructing a student academic performance matrix ,in, Indicates the first Period, No. Standardized average academic performance of students taught by each teacher This indicates the final grade. Indicates basic entrance examination scores. This represents the total number of students; For the first Period, No. Teacher, No. All raw score data for each indicator Calculate its mean and standard deviation If the score of a questionnaire meets the requirements If a value is found to be an outlier, it will be removed, thus completing the data cleaning process. The cleaned data was integrated into a related dataset. , This represents the original score matrix of teaching evaluations after cleaning. This represents the cleaned student academic performance matrix.

3. The comprehensive evaluation method for teaching quality based on the credibility of teaching evaluation data according to claim 1, characterized in that, In step S2: For the Period, No. The teacher, obtain all of them The homogeneity coefficient of the scores of each evaluation indicator is calculated using the following formula: ， in, Indicates the first Period, No. The homogeneity coefficient of the evaluation data of individual teachers. Indicates the first The teacher in the The standard deviation of all indicator scores during the period Indicates the first The teacher in the The arithmetic mean of the scores of all indicators during the period.

4. The comprehensive evaluation method for teaching quality based on the credibility of teaching evaluation data according to claim 3, characterized in that, In step S2: a homogenization correction factor is generated based on the homogenization coefficient, a preset homogenization threshold, and the total number of valid questionnaires, specifically as follows: ， in, Indicates the first Period, No. Homogeneity correction factor for teachers Represents the homogenization threshold, the first Period, No. The total number of valid questionnaires for each teacher is calculated by summing the number of valid questionnaires for all evaluation indicators for that teacher.

5. The comprehensive evaluation method for teaching quality based on the credibility of teaching evaluation data according to claim 4, characterized in that, In step S2: Based on the original evaluation weights and original scores, and combined with a homogenization correction factor, the corrected individual evaluation scores are obtained. ， in, Indicates the first Period, No. The preliminary composite scores of the teachers after homogenization correction Indicates the first The preset weights of each evaluation indicator and , Indicates the first Period, No. Teacher, No. The original average score of each indicator Indicates the first Period, No. Homogeneity correction factor for teachers.

6. The comprehensive evaluation method for teaching quality based on the credibility of teaching evaluation data according to claim 1, characterized in that, In step S3: Utilizing history Period data, for the first period Each evaluation indicator was used to obtain its score and the correlation coefficient between the score and academic performance. ， in, Indicates the first The correlation coefficient between the scores of each evaluation indicator and academic performance scores - grades. Indicates the first Period, No. Teacher, No. The original average score of each indicator Indicates the first Period, No. Standardized average academic performance of students taught by each teacher Indicates the first Historical average scores of each indicator This represents the historical average academic performance.

7. The comprehensive evaluation method for teaching quality based on the credibility of teaching evaluation data according to claim 6, characterized in that, In step S3: The score-performance correlation coefficient is compared with a preset correlation coefficient threshold, and the adjusted unnormalized weights are obtained by combining the internal consistency reliability coefficient: ， in, Indicates the first The adjusted unnormalized weights of each evaluation indicator Indicates the first The preset weights of each evaluation indicator, This represents the correlation coefficient threshold. Indicates the first Internal consistency reliability coefficient of each evaluation indicator; Among them, the The internal consistency reliability coefficient of each evaluation indicator is expressed as: ， in, Indicates the first The internal consistency reliability coefficient of the i-th evaluation index represents the... The total number of scoring items under each evaluation indicator This indicates the number of all historical cycles, all teachers, and all questionnaires. The variance is calculated by combining the scores of each rating item into a single sample. This indicates that all historical cycles, all teachers, and all questionnaires will be included in the first... The variance is calculated by combining the total scores of each evaluation indicator into a single sample.

8. The comprehensive evaluation method for teaching quality based on the credibility of teaching evaluation data according to claim 7, characterized in that, In step S3: The unnormalized weights of all evaluation indicators are normalized to obtain the corrected weights: ， in, Indicates the first Adjusted weights for each evaluation indicator This is the sum of the unnormalized weights of all indicators.

9. The comprehensive evaluation method for teaching quality based on the credibility of teaching evaluation data according to claim 1, characterized in that, The specific steps of S4 are as follows: Based on the latest evaluation cycle The original evaluation score, the first The final weights of the evaluation indicators after adjustment, along with the homogenization adjustment factor for the latest period, are used to obtain the [number]th evaluation indicator. The final overall evaluation score of each teacher: ， in, Indicates the first The final comprehensive evaluation score of each teacher, Indicates the latest evaluation period , No. Teacher, No. The original average score of each indicator Indicates the first Adjusted weights for each evaluation indicator Indicates the first Homogenization correction factor for each teacher in the latest evaluation cycle.

10. The comprehensive evaluation method for teaching quality based on the credibility of teaching evaluation data according to claim 1, characterized in that, The corrected weights calculated in step S3 are used to update the weight system for the next evaluation cycle, forming a dynamic iterative optimization closed loop for the evaluation index weights.

Images