Study On Influencing Factors And Prediction Model Of Doctoral Dissertation

The study and research of Ph.D.students in the education phase is no longer a simple process of knowledge acceptance and accumulation,but the innovation of knowledge through exploration and research.The "Provisional Measures for the Implementation of the Degrees Regulations of the People's Republic of China" pointed out that doctoral students should have the ability to independently engage in scientific research as well as innovation in science and technology.Therefore,the quality of doctoral dissertation is the most basic and most central indicator in the quality of doctoral student training.Therefore,the influencing factors and prediction research on the quality of doctoral dissertation have great significance for students,teachers and schools,and provide an objective basis for improving the quality of doctoral dissertation.At present,there are many researches on the quality of dissertations at universities both at home and abroad.Generally speaking,it mainly focuses on two aspects:assessment of regulatory and influencing factors.However,research methods mainly focus on descriptive statistics.Dr.is the pinnacle of international formal education qualifications and represents excellence.Therefore,it is necessary to study the influencing factors and forecasting quality of doctoral dissertation.This thesis mainly analyzes the materials of the examination and approval of doctoral dissertation,evaluation results and questionnaire survey data held by a graduate school of a university to explore the influencing factors of the quality of doctoral dissertation,which mainly includes four research contents:(1)Overall analysis of dissertation quality.The descriptive statistics are used to analyze the status quo of the quality of doctoral dissertations in this university from three aspects:graduation year,evaluation index,and cultivation unit;(2)Analysis of influencing factors on quality of doctoral dissertation.Starting from the nine factors of the source characteristics and training methods,the thesis explores the related factors that influence the quality of doctoral dissertations,and analyzes the data of the questionnaires to improve the factors that influence the quality of doctoral dissertations;(3)Research on prediction model.Taking the important influencing factors obtained from the correlation analysis of the previous section as features,support vector machine(SVM)and random forest(RF)algorithm are used to model,and on this basis,a dual-weighted random forest algorithm is proposed,and finally the prediction results are compared.Obtain the optimal model;(4)Based on the research results,put forward the corresponding training suggestions for doctoral candidates from the doctorate itself,the tutor,and the school.Through the research,the main conclusions are as follows:(1)"Results innovation and benefits" is the weakest of the five secondary indicators.The low innovation is a common problem in domestic colleges and universities,and it is also important for colleges and universities to pay attention in the future.(2)Enrollment age,study period,mode of study,subject categories,whether cross-specialty and type of thesis research have a significant impact on the quality of doctoral dissertations.The higher the enrollment age and the longer the study period,the lower the average evaluation value of doctoral dissertation quality;The quality of doctoral dissertations was high when the mode of study is successive master-doctor program,the subject categories is science,and type of thesis research is basic research,whether cross-specialty is no cross-major.(3)The double-weighted random forest method achieves a prediction rate of 81.29%,which is better than the traditional RF and SVM.The dual-weighted method was used to reconstruct the random forest model,which overcomes the influence of sample data imbalance on the prediction results,enhances the generalization ability of the random forest model,and improves the overall prediction accuracy of the model.The research provides an objective basis for the quality management of doctoral dissertations and the cultivation of doctoral students in universities.