Design Framework of Early Warning Mechanism of Student Achievement with Machine Learning

With ever-changing educational circumstances, teaching management methods are constantly innovated. Especially, the teaching management practice has produced a large amount of student data information. The students’ data covers a wide range of contents, including students’ daily test results, students’ rewards, and punishments records as well as basic personal information. Thus, how to achieve the accurate screening and efficient use of valuable data information is worth discussing. At the present stage, the existing technologies and conditions have used these data fully. Therefore, by finding the methods for the scientific use of student data and mining the information value, we can better guide students to learn, strengthen student management, and help students improve their grades with positive supervision and early warning. In this context, based on a brief review of related theories and techniques, this study developed an early warning mechanism for student performance based on machine learning. This paper summarizes the theories and technologies related to educational data mining and proposes the design of the student performance early warning system based on machine learning according to the system requirements analysis, overall architecture design, multi-source heterogeneous data acquisition, and preprocessing.