Identifying Key Features in Student Grade Prediction

Abstract

With the development of education data mining and the data of academic affairs accumulated, the performance of students in school could be analyzed from different views and explore more precious aspects which influence the grades of students. Our research conducts data mining on student basic courses information, learning behavior information and admission information, which will help to find the relationship between them. This work mainly focus on exploring the key features that take the important roles in student academic performance. Then the work takes the consider of identifying the relationship between student behaviors and their grades. By using the advanced machine learning methods and feature analysis methods, LASSO, the work rated the most important features of student behaviors. We found several key relationships between student behaviors and their grades, for example, the more books one borrows, the better grade he/she will get. This work would help the educators and students to better understand the relationship between connotative factors and the student achievement.