Student score prediction method based on campus big data

Abstract

The invention relates to the field of education data mining, and provides a student score prediction method based on campus big data, which comprises the following steps: (1) data fusion; in a smart campus environment, dynamic and non-invasive data acquisition is carried out by taking learners as centers, so that convergence and fusion of online and offline and in-class and out-of-class multi-dimensional behavior data of students are realized; (2) feature calculation; by comprehensively utilizing linear analysis, deep learning and nonlinear analysis technologies, a student multi-dimensional behavior characteristic system is systematically constructed from three aspects of behavior change, diligence and behavior nonlinearity, and a student behavior mode and dynamic change thereof are deeplymined; and (3) score prediction; based on a machine learning algorithm, a high-precision score prediction model is constructed, and feedback and early warning are provided for high-risk student groups. According to the method, student scores can be scientifically and comprehensively predicted, and early warning is provided for students with high hanging risk.

Description

technical field [0001] The invention relates to the field of educational data mining, in particular to a student performance prediction method based on campus big data. [0002] technical background [0003] The rise of the Internet of Things, big data and artificial intelligence technology, on the one hand, provides a more powerful development environment for the construction of a smart campus that is currently being explored; on the other hand, while providing convenience for teachers and students, it also accumulates Considerable massive data resources. Mining the rich information behind these data will not only help to deeply understand the behavior patterns of students, but also has important research value for exploring personalized learning services; it will also help to evaluate the current development status of smart campuses, and contribute to the development of smart campuses. It has important reference value for further construction and optimization. [0004] Fo...

AI Technical Summary

Problems solved by technology

However, previous studies on student performance prediction have not fully utilized the natural advantages of this massive data, and the related studies have the following shortcomings: previous studies either relied on questionnaires, which was time-consuming and labor-intensive, and the efficiency was not high, and the feasibility was also limited It is worthy of scrutiny; either only focusing on a single student behavior (for example: online learning), cannot fully reflect the student's behavior pattern; or does not consider the influence of time factors, and cannot reflect the dynamic change process of the student behavior pattern over time

Images