Research And Application Of Needy Students Identification Model Based On TabNet-Stacking

Most colleges and universities currently rely on manual review to identify needy students,that is,student application,class review,counselor review,and publicity of the results.This traditional identification method relies on the student’s application materials,and the review process is subject to a certain degree of subjectivity.Some needy students will not apply for funding because of the shame of publicizing the results.These problems will lead to inaccurate identification of needy students,which is not conducive to the development of identification of needy students and affects the distribution of various scholarships and grants.Therefore,it is of great practical significance to carry out the work of needy students reasonably and efficiently and improve the accuracy of the identification results for student management and education fairness.With the construction of the university’s informatization business system and the improvement of the data platform,the data generated by the students in the school can be completely preserved.The data mining method is used to analyze various data of the students and obtain the basis for the identification of the needy students to realize the identification of the needy students.Work is an effective way.The main work of this paper is to collect the data of students in school,use the data mining algorithm to design the identification model of needy students,and realize the identification of needy students by constructing the identification and identification system of needy students.There are three specific points:(1)To carry out the project on student behavior characteristics.This paper collects data sets related to students through various business systems of colleges and universities,combines the actual work experience of needy students and the method of feature selection,and constructs a set of students’ behavior characteristics from four dimensions of students’ basic information,academic situation,consumption situation,and diligence to reflect student status at school comprehensively.(2)Design the TabNet-Stacking identification model for needy students.In this paper,the TabNet model is used to achieve the important ranking of student behavior characteristics.According to the multi-source characteristics of student behavior data,the Stacking integrated learning method is used to construct a classification model with strong generalization ability and high accuracy.The student behavior data is analyzed to obtain student categories and realize poverty.Identification.(3)Design and implement the identification system for needy students.This paper analyzes the need for the identification and identification system for needy students and confirms the system architecture design scheme and system function modules through business requirements,user requirements,and functional requirements.We are using Think PHP6 and Bootstrap to realize the development of needy students identification systems and verify the effectiveness of the model in practical work.The experimental results show that the TabNet feature engineering proposed in this paper combined with the Stacking ensemble learning classification method for identifying needy students is more efficient and accurate on the student behavior data set of a university in the east than other needy student identification models,and the F1 value performance is better than The other models averaged 1.61% higher.Using the identification model of impoverished students proposed in this paper,the identification and identification system of impoverished students constructed can provide a better decision-making role for the work of impoverished students in colleges and universities.