Research And System Implementation Of Achievement Early Warning Based On College Student Data

With the continuous development of big data in education,more and more researchers have invested in this research field.As an important branch in this field,achievement prediction has become a hot spot for researchers.In this paper,features are extracted from the data of students in school,a achievement prediction model is established,and on this basis,a achievement early warning system is designed and realized to help counselors better manage students' learning and help students successfully complete their studies.The specific work contents are as follows:(1)This paper uses ETL tools to collect student card consumption data,library access control data,library borrowing data,campus network log data,and extract student behavior characteristics from these data,such as the number of days in the library during the week,the number of book borrowings,and consumption The number of times,breakfast regularity,etc.,the correlation analysis of these extracted features and the student's achievement,to find out the features that affect the student's achievement.(2)This paper builds a achievement prediction model based on the Stacking integrated learning method.Because the data is unbalanced,the Smote algorithm is used to oversample the data.In order to reduce redundant features and improve the classification performance of the model,the TMGWO algorithm is used to perform feature selection on the data.The original TMGWO algorithm uses accuracy as a fitness function.For the evaluation of unbalanced data,this paper improves the algorithm and uses F1-Score As the fitness function of the TMGWO algorithm,it evaluates the selected features and verifies the effectiveness of the improvement through comparative experiments.Finally,the achievement prediction model constructed in this paper has an accuracy of 87% and an F1 value of 0.8 on the test set.(3)This paper designs and implements a performance early warning system based on the constructed achievement prediction model.The system is based on the B/S architecture and uses front-end and back-end separation technology.The main users of the system are counselors and students.Counselors can view the early warning information and historical results of students through this system.When a student is warned,the counselor can send emails to students to remind students through the system.Through this system,students can view their own grade warnings and historical grade information.