Research And Implementation Of Academic Early Warning Model Based On TabNet

Education is the foundation of our country,and it is the basis for our country to move from a large human resource country to a strong human resource country,and higher education is the top priority of education.The number of students with learning difficulties has also increased,so it is very important to provide academic support to students with learning difficulties and help them to finish their studies better.However,the existing academic early warning methods in universities use the total credits of failed courses to warn students academically,which has the problems of long warning period and lag of students receiving warning signals.(1)To address the problems of long warning cycle and warning lag,this paper combines student behavior data and historical performance data to investigate their correlation with academics,shortens the warning cycle to three months,and one month before the final exam can issue a warning notice to students with learning difficulties.The correlation between student characteristics and academics is analyzed using frequency statistics and Kendall rank correlation coefficient analysis method,and the student characteristics with high correlation are used as the input features of the early warning model.(2)A KNN-Tab Net-based academic early warning model is proposed.To address the problem of poor performance of Tab Net native missing-fill method in practical applications with small data size,we propose to use KNN interpolation fill method instead of Tab Net missing-fill method to improve the classification performance of the model,and then use Bayesian algorithm to tune the parameters of Tab Net hyperparameter space to accelerate the convergence speed of the model,and finally compare the KNN-Tab Net early warning model proposed in this paper with Finally,the proposed KNN-Tab Net early warning model is compared with Tab Net,XGBoost,Light GBM and other algorithmic models,and the performance of the proposed algorithmic model is verified to be better.(3)Design and implementation of KNN-Tab Net-based academic early warning system.Firstly,we analyze the functional and non-functional requirements of the academic warning system,then design the architecture and database of the academic warning system,then use Think PHP and Admin LTE framework to build the academic warning system,and embed the KNN-Tab Net model proposed in this paper into the academic warning function module,and finally use the black box testing method to test the stability and ease of use of the system.Finally,the stability and ease of use of the system are tested using black-box testing methods.