Research On The Reliability Method Of University Students’ Evaluation Of Teaching Based On Data Mining

Students’ evaluation of teaching is an important part of teaching evaluation in colleges and universities.The quality of students’ evaluation of teaching directly affects the accuracy of teacher evaluation,which in turn affects all aspects of teaching activities.At present,the research on data of teaching evaluation in colleges and universities mainly focused on the research of indicators of teaching evaluation and the correction of data of student evaluation.The above problems have not been solved from the attitude of students to evaluate teaching,which has certain limitations.In this study,according to the data of students’ teaching evaluation in Hebei Agricultural University,a set of feature sets is constructed to evaluate the students’ teaching evaluation reliability,and then the number of students’ reliability grades is reasonably determined by using the clustering algorithm.Finally,the classification algorithm is used to construct the prediction model of students’ teaching evaluation reliability.Through the feedback of teaching evaluation grade in the form of star,we can strengthen students’ awareness of teaching evaluation,promote students’ correct attitude of teaching evaluation,make students’ teaching evaluation more effective and objective,so as to make teachers’ evaluation results more fair,and make university teaching evaluation better provide strong support for school decision-making.The main work of the thesis is as follows:(1)Construction of reliability features of teaching evaluation data.This paper collects the data of students’ teaching evaluation in Hebei Agricultural University.Through the exploratory analysis of the objective scores and subjective comments in the teaching evaluation data,a group of data features that can effectively measure the reliability of students’ teaching evaluation are constructed.At the same time,the normalization method is used to unify the dimensions,and the correlation coefficient analysis is carried out on the data features,so as to further optimize the data features and provide reference for teaching evaluation.(2)Reliability grading of college students’ evaluation of teaching based on improved Kmeans algorithm.According to the features of student reliability,combined with the Kmeans algorithm,a K-means algorithm based on neighborhood spatial density optimization is proposed.The algorithm first divides the data into grids to filter out high-density networks.Then the algorithm continuously merges the grid centers according to the iteration factor to obtain a candidate grid set.Finally,the algorithm selects the initial clustering center based on the distance algorithm,and then performs K-means clustering.The accuracy of the improved algorithm on the public data set has increased by more than 10%,and the iteration efficiency has improved.The improved k-means algorithm is used to cluster the students’data,which solves the problem of students’ reliability classification,realizes the annotation of students’ reliability data.(3)Prediction of teaching evaluation reliability based on improved BP neural network.According to the results of student reliability features and data annotations,this paper proposes an improved BP neural network classification algorithm based on Crow search algorithm to build a model for student evaluation of teaching reliability.The algorithm first determines the network structure and parameters of the BP neural network based on experiments.Then it uses the crow search algorithm to optimize the initial threshold and weights of the BP neural network,avoiding the result from falling into the local optimum,and improving the prediction accuracy.The improved algorithm has a prediction accuracy rate of 95%on the student evaluation data set,which is 5%higher than the accuracy of BP network,and the results have better stability.