Research On College Students' Behavior Analysis And Prediction Based On Campus Card Data

The rapid development of social economy,the intelligent management of colleges and universities has received more and more attention.The times are improving,and we must also follow the progress of our university management to better serve our students.Now entering the era of big data,all walks of life will produce a lot of data.The school forms all kinds of information all the time,such as the daily consumption record of the card,the family information of the students,the basic information of the individual and so on,various types of data can reflect the daily life of students.As the university database is superimposed over time,its data accumulates more and more,which together constitute a huge large data environment.Based on this,it is very necessary to analyze the data and dig deep into the value of the data.This article mainly uses Cluster analysis and big data technology,based on various information formed by students during the school,characterize student portraits based on students' academic and daily life,and analyze student behavior,so that managers can accurately and comprehensively locate student groups and pay attention to students.To lay the foundation for refined management.The main research contents are as follows:(1)The data of all aspects of university students are collated and combined with data to design student evaluation indicators.Designing a subdivision model based on cluster analysis,the initial clustering center selection and distance calculation of the k-means algorithm are improved,and an improved k-means clustering algorithm is proposed to make the clustering effect better and to classify students accurately.(2)In view of the post-processing and timeliness of student behavior reminders,this paper proposes a K-nearest neighbor algorithm based on student behavior categories to predict student behavior,allowing counselors and teachers to manage and help students in a timely manner.Realized the change of student behavior post-emergency emergency to pre-existing early warning guidance,guiding students to the right track.This paper established a student behavior feature category model,enhanced the prediction model of the K-nearest neighbor algorithm,and improved the prediction of student behavior accuracy.(3)This paper studies the relevant theories based on the Support Vector Machine(SVM)algorithm for the special groups of college students,combining student consumption to determine economically impoverished students,and better help schools to judge poor students.Then combined with the student's relevant information to make predictions about whether the students will have abnormal behavior,promptly let the teacher interfere with the students and help the students out of the predicament.By studying these contents,we can provide better help for student management staff and improve the self-work quality,service quality and work efficiency of student management staff,especially improve the understanding of managers on students.Therefore,it helps university administrators to take effective measures to manage students in a targeted manner,and truly practice teaching students in accordance with their aptitude and assisting in the generation of mechanical materials.