Design And Implementation Of Graduate Admissions Information Analysis And Forecast System Based On Historical Data

The quantity and quality of postgraduate students are one of the most important issues for the construction and development of college subjects.Adequate use and effective management of student enrollment information can provide the basis for making decisions for the successful completion of enrollment.Among them,the forecast of Graduate Student Scoreline,the statistics of Student Source and the prediction of professional enrollment results are the problems that managers are very concerned about,which helps schools to take effective response measures in advance.The thesis aims at the design and development of graduate student enrollment information analysis and forecasting system,which focuses on the prediction methods for national admission scores,and achieves predictions,information management and statistical analysis,designs and development of functions such as visualization.For the prediction problem of national score for each discipline,we propose a method which bases on grey relational analysis combined with multiple linear regression for prediction,and add all the factors that may be considered that affect the scoreline to the prediction model,to make predictions closer to real scores.In this method,an automatic threshold selection method is designed to obtain an independent variable whose correlation is greater than the threshold,and then solving the possible collinear problem of these selected independent variables.Finally,using multiple linear regression method to predict the score line.The thesis analyzed the actual business processes and requirements from the entry of candidates to admissions,and completes system function design based on B/S architecture.The thesis adopted MVC mode,ASP.NET and SQLServer database to implement system development.The system which is based on the network management of candidates entering the admissions process designes general management functions,adds a national score forecast,student source analysis,related statistics,visualization and other functions.The system uses the enrollment data of a university for 2012-2018 to conduct experiments and forecasting national admission scores for 2017 and 2018 respectively,then compared with real scores.Experiments show that the prediction model proposed in this thesis whis error is only about 1%.The result has a good effect on the prediction of scores of various disciplines.Simultaneously,compared with a linear regression equation,multiple linearregression and evaluation forecast model,the result is clearly better than other models.