Research on application of early warning of students achievement based on data mining

Abstract:

Academic achievement is the most important and most basic indicator for evaluating a student's learning situation. The analysis of academic performance is beneficial to teachers to master a student's learning situation and conduct the targeted teaching and counseling, while for the students, knowing in advance what happens to their future learning is also beneficial for students to discover their own problems which could be prevented in advance. Most of the existing research work is based on the analysis of the curriculum, historical performance or behavioral data to predict the student's total score, while few studies focus on combining a student behavior with his or her grades to comprehensively predict a student’s learning in all future courses. This paper, from a new perspective, uses the three aspects data of student behavior, personal attributes and historical achievements which are identified by the influencing factors based on students' different curriculum dynamics in the future, the early warning of students' achievement is predicted with support vector machine, the experiment results show the exploratory work of data mining applied in education has certain significant meaning for the teachers as well as for students.