Font Size: a A A

Research On Student Performance Evaluation Based On Random Forest Algorithm

Posted on:2018-05-04Degree:MasterType:Thesis
Country:ChinaCandidate:L Q ZhuFull Text:PDF
GTID:2347330518984074Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
Recently,audit evaluation on undergraduate teaching and professional certification for engineering education are very important work in the higher education sector.The core of the work is to evaluate the goal and the effect of talent cultivation.An important part of the audit evaluation is to evaluate the effect of personnel training.Colleges and universities are mainly to carry out effective assessment of students through the examination.Therefore,the student's test scores are the most important indicators to measure the effect of college students on the training.In some extent,the evaluation of the students in school is to evaluate their academic performance.Evaluation of student performance is not only an important part of school routine management,but also an important part of teaching work.Scientific and reasonable evaluation of student achievement,on the one hand,can make students more comprehensive and in-depth understanding of their learning situation.n the other hand,it allows teachers to quickly and accurately grasp the learning situation of students.It can also promote the implementation of scholarship assessment,postgraduate deduction and provide decision-making basis for the scientific teaching management.The traditional method of evaluation on student achievement mainly conducts a simple treatment for student performance,teaching staff cannot obtain targeted information for improving the future work from the performance data.Therefore,a comprehensive and scientific assessment of student achievement in the assessment of education and teaching work is particularly critical and urgent.Random Forest is a kind of combined self-learning technology.Random forest can classify the importance of features in the classification.Random forest has a good anti-noise ability.Two randomness is introduced,so it is difficult fall into over-fitting.Based on the random forest algorithm,this thesis presents an algorithm named ESP_RF algorithm which is for scientific evaluation of student achievement.ESP_RF algorithm is used to predict the relevant grades of the students in senior grades by the students' junior course grade.It can also be used to sort out the factors that affect student performance.The ESP_RF algorithm is used to evaluate the student achievement in a university in Anhui.The results show that the model can be a good student performance prediction and the order of importance of the characteristic variables is consistent with the actual situation.Using scientific methods to predict student achievement not only can guide students to have targeted tutorials,but also provide scientific decision-making basis for the teaching management personnel.
Keywords/Search Tags:performance evaluation, random forest, feature importance, prediction, ESP_RF algorithm
PDF Full Text Request
Related items