Research On Student Portrait And Academic Early Warning Method Based On Data Mining

With the rapid development of the Internet of Things,big data and artificial intelligence technologies,the level of informatization in various industries has been continuously improved.The development of informatization provides important conditions for the reform of education,and exploring smart classrooms and building smart campuses has become an inevitable way for school reform."China's Education Modernization 2035" pointed out that combining big data,machine learning and other technologies to build a digital and intelligent smart campus platform to realize smart education,smart management and smart services is the inevitable trend of the current education informatization development.In the process of smart campus construction,many colleges and universities still use the traditional student management method,which simply divides students according to majors,grades and other conditions,and lacks personalized management and services for students' behavioral characteristics.This dissertation applies machine learning technology to the study of student portraits and academic early warning,and builds student data portraits and academic early warning models to help university education administrators fully understand students,accurately predict students and provide personalized services to students.The main research contents and contributions are as follows:(1)Aiming at the problem that the clustering results tend to fall into the local optimum when the K-means algorithm constructs the student portrait model,a K-means algorithm based on Canopy and the principle of max-min distance is proposed.The improved K-means algorithm is used to cluster and analyze the student data from the two dimensions of consumption and learning,and the word cloud graph method is used to construct student portraits according to the clustering results.The experimental results show that the improved K-means algorithm can effectively distinguish students with different behavioral characteristics and help schools to understand students more comprehensively.(2)Aiming at the problem that the SVM algorithm uses the default hyperparameters to build an academic early warning model,the prediction model cannot achieve the ideal accuracy rate,an SVM algorithm based on improved FOA is proposed.Based on the historical performance data of students in the first three years of a university,the consumption data of Campus All-in-one Card and the data of library borrowing and access control,the SVM algorithm based on improved FOA is used to predict whether students can graduate smoothly in the future,and send an early warning to students who may not graduate smoothly in the future.Experiments show that the SVM early warning model based on the improved FOA is better than the traditional support vector machine,decision tree and random forest three classification models in terms of accuracy.(3)Aiming at the campus data set of a university in Henan,a student behavior analysis system based on data mining is designed.The system is divided into four modules,namely login module,data processing module,student portrait module and academic warning module.Experiments and tests have been conducted on the system,and the results show that the system can visualize the behavioral characteristics of school students,and can help school administrators personalize their services and management of students,which has certain practical value.