College Students Learning Behavior Analysis Based on SVM and Fisher-Score Feature Selection

Abstract

With the development of modern educational methods, new modes and platforms for college students’ learning have appeared. Through the scientific statistics and analysis of students’ learning behaviors, we can find the regular pattern contained in these data, discover students’ interest goals and predict learning effects, and provide students with targeted and personalized learning guidance. At present, the common analysis methods are mainly K-means clustering method. Considering the support vector machine has higher classification accuracy, this paper proposes a support vector machine analysis method based on Fisher-Score feature selection for students learning behavior analysis. Firstly, through the Fisher-Score feature selection, the key features in the learning behavior are selected, and the honor features unrelated to the learning effect are removed, and then the data analysis is performed by the SVM classifier. The verification results show that our method has better accuracy.