Capability Indicators Analysis And Graduation Forecast Based On Data Mining

With the deepening of higher education reform and the seriousness of the real social employment situation,graduates do not have a correct understanding and estimation of their real abilities,so they are hesitant about the choice of graduation destination after graduation.In fact,most universities have employment centers which provide students with career counseling services,but currently,there is no suitable reference standard,and considering the severe employment situation and the difficulties of higher education reform,obviously,it is hard to achieve personalized guidance.A student's ability normally can be assessed in terms of comprehensive performance,foreign language proficiency,and practical activities,but if there is an ability indicator to evaluate students' different levels of competence,students can more objectively understand their abilities and make a correct judgement on their graduation direction.This paper focuses on students majoring in computer science and technology.According to the students' comprehensive performance,basic information,graduation information and other data,the relevant competency indicators are generated,and data mining techniques are used according to the competency indicators and certain school behavior information to analyze and forecast students' graduation destination,which processes mainly from three aspects:(1)Using k-means clustering algorithm,the students of computer science and technology are reasonably divided into several clusters to analyze the overall graduation trend of students,compare the differences of each ability index of each cluster students and research the relationship between the students of different clusters and the graduation destination.(2)Using the Apriori association rule mining algorithm to explore the relationship rules between student ability indicators and graduation destinations,as well as the relationship rules between student behavior information and ability indicators and to analyze the most influential factors affecting students' graduation destination.(3)Five kinds of machine learning algorithms,such as Random Forest,SVM,LR,GBDT and Neural Network,are used to establish the graduated prediction model.The model evaluation indicators are used to compare and evaluate the results,and the most appropriate prediction algorithm will be selected to predict the student's graduation destination.Besides,the importance of various characteristic factors will be compared.From the above research experiments,through the clustering results analysis,different students of different clusters are selected for the different choices of graduation destinations,and those students with excellent ability indicators are more motivated to graduate.What are the capabilities analyzed through the association rule mining algorithm? The relationship between indicators and school behavior information and graduation destination through the establishment of graduation direction prediction model and the comparison of model evaluation index results,the best model based on ability value prediction is logistic regression and ability indicators.The best model is SVM.In addition,various characteristic factors are predicted as individual characteristics.It is discovered that the ability index has a greater impact on graduation and the gender has the least impact on graduation.