Analysis Of Student Performance In School And Application Research Based On Machine Learning

Education has always had a profound impact on economic and social development,which not only affects and determines the individual development of people,but also affects and determines the future development of human society.All countries in the world attach great importance to education and attach importance to the cultivation of talents.Talent cultivation is the primary function of universities and one of the main goals of education.Undergraduate students are the main object of college education,and strengthening undergraduate education is the main task of college education.How to understand and grasp the learning and academic situation of students,and make reasonable and appropriate evaluations of students? In the past,students’ academic performance was usually based on academic performance,so it was easy to ignore the influence of other factors.Therefore,adhering to the concept of teaching according to aptitude,colleges and universities should put forward plans to adjust and improve teaching strategies in combination with academic conditions,so as to improve students’ learning ability and comprehensive quality.This paper takes students as the first person,uses the form of online questionnaires to obtain students’ learning data,uses machine learning algorithms to construct students’ comprehensive evaluation level prediction model,and evaluates them,taking into account other factors that affect students’ learning conditions,avoiding the influence of achievement factors,making evaluation more objective,and better promoting the improvement of talent training quality.The research in this paper mainly includes the following aspects: First,through the introduction of the research background,the purpose and significance of this paper are introduced.Secondly,the relevant theories of classification trees,BP neural networks and convolutional neural networks are expounded,which paves the way for subsequent empirical evidence.Third,through data preprocessing（deletion of missing values）,multiple collinearity test,construction of index system and other steps,the classification tree model,BP neural network model and convolutional neural network model of students’ comprehensive evaluation level are established.Due to the imbalance of samples in the dataset,in order to improve the prediction accuracy of the model,this paper adopts sampling methods such as original sampling,oversampling,undersampling and MOVE sampling for comparative analysis.In this paper,on the basis of the standard BP neural network,the number of input and output channels of the hidden layer is adjusted to obtain the optimal model;since the convolutional neural network is more used in image classification and less used in the field of academic analysis,this paper adopts the convolutional neural network consistent with the optimization method of BP neural network,and compares and analyzes with other two models.Finally,the output importance indicator system.Through the analysis of the model output results,this paper draws the following conclusions: first,in terms of model prediction accuracy,the prediction accuracy of the classification tree on the comprehensive evaluation level of students is higher than that of BP neural network and convolutional neural network;secondly,in terms of model running time,the classification tree model is better than the BP neural network model and the convolutional neural network model.Combined with the model prediction accuracy and running time,the classification tree performance under THE SAMPLE sample is satisfactory.In terms of the importance of indicators,after analysis and comparison,it is concluded that X1-average weekly learning time inside and outside the classroom,X9-explaining to students,X4-average weekly leisure and entertainment time,X7-participation in internships,the number of social practice activities,X62-stress relief methods,X63-talent training positioning,X2-average weekly extracurricular activity time,X5-the number of reports or speeches,X66-mother’s education level,X64-place of origin and other indicators play the most important role in the prediction of students’ comprehensive evaluation level.Comprehensive analysis and research,this paper puts forward the following suggestions: for schools,it is necessary to collect the factors affecting students’ learning conditions in a timely manner,especially the important characteristics and indicators,so that schools can grasp the learning conditions in a timely manner,track the quality of students’ learning and adjust the education and teaching strategies in a timely manner;for students,according to their own situation,strengthen learning and training according to the evaluation indicators,and use the information to improve learning efficiency,so as to improve their own learning ability and comprehensive quality.In summary,the research of this paper has certain theoretical significance and practical value in predicting the classification of students’ comprehensive evaluation levels and improving the quality of talent training.