Research on teaching state space dimensionality reduction and enhancement strategies incorporating principal component analysis in open education platforms

Abstract

In the context of the rise of online education, learners have generated large-scale learning behavior records in open education platforms providing sufficient material for educational data mining. In this paper, based on edX open education dataset, 11 features representing learners' learning status are downscaled by principal component analysis technique, learners are clustered based on learning status and using the proposed K-means clustering algorithm, and the number of clusters is determined by the aggregation coefficient method. The experimental results show that the aggregation coefficient peaks when the number of clusters is 4. Therefore, the optimal teaching state is categorized into 4 types of spaces, which are general learning space, negative learning space, interactive learning space, and active learning space. Finally, on the basis of the clustering results, corresponding teaching enhancement strategies are proposed for the strengths and weaknesses of the groups. The method proposed in the article can identify and analyze students' learning status in real time and accurately, improve teachers' mastery of students' classroom performance, and provide a new research idea and technical reserve for the development of open education.