Student Behavior Analysis Based On Big Data Technology

With the development of the times,it has accelerated the informatization of student data in the campus and made the student management work of the school more efficient.Various management departments in the school store more student data on the computer.With the accumulation of data,more and more data are collected from relevant departments.These data are closely related to students' behaviors at school,which can well reflect students' behavior habits and learning status.Thus,the use of data mining technology to observe and analyze student behavior data is very important for the management and construction of schools and students,and can improve the efficiency of school management.This topic uses the campus card consumption data of a university in hangzhou and the list of financial aid from the office of academic affairs to carry on the research on the classification and prediction of poor students,and uses the online log of a university in zhengzhou to cluster the students' online behaviors.School has begun to widely use the campus one-card system,students in the daily consumption process produced a large number of one-card data;Taking the relevant data generated by students in the one-card service center and information center as the research object,the data mining technology is applied to the consumption data of campus one-card for in-depth mining.Based on the machine learning classification algorithm,this paper constructs the classification model of poor students in colleges and universities,and puts forward the PS-XGBOOST model that combines PCA and SMOTE algorithm to predict,which provides an important basis for the evaluation standard of poor students in colleges and universities in China,and guarantees the impartiality of the identification of poor students in colleges and universities.On the basis of deep learning K-MEANS algorithm and CANOPY algorithm,the parallelization algorithm of these two algorithms was designed based on HADOOP platform.At the same time,the improved CAK-MEANS algorithm was proposed,and the improved parallel algorithm was deployed in the HADOOP cluster composed of three machines to study the online data of students in school.Through combining big data technology and clustering algorithm,this topic analyzes the behavioral data ofstudents from all dimensions,and the results obtained can help teachers have a further understanding of students.Moreover,the use of big data technology to analyze the behavior data of students can also provide decision-making help for the management of campus network from a technical perspective.