Online Learning Behavior Analysis Based On Behavior Log Data

With the development of science and technology,online education has gradually matured.In the course of its development,participants in online learning have produced a huge amount of learning behavior data.For educators,these data are undoubtedly of great research value.The exploration,analysis and mining of these data will help the learners to learn and summarize the learning characteristics and behavioral rules,thus providing guidance for the improvement of instructional design and adaptive learning of students.This paper reviews the development history of MOOC,sorts out the research status and research characteristics at home and abroad,and based on the research done by the predecessors,combined with the current development status of MOOC,selects the well-known domestic MOOC platform-Xuetang Online-the course study The data was used as the original data for the study.This study focuses on the use of a variety of data analysis tools and research methods to extract,clean,normalize and statistically analyze the behavior log data selected for the study on the platform.The main concern is the selection of the timeline of the platform course,the learner's various extracurricular learning behaviors during the course,the learner's learning time and the frequency of learning in the learning cycle.The paper attempts to use data visualization technology to analyze the relationship between the start time of the course and the enthusiasm of students,to find the time points for students' preference for online learning,and to explore the enthusiasm of students for learning between different behavioral dimensions by means of correlation analysis and independent sample test.influences.The K-Means clustering algorithm is used to divide the MOOC learners into four types.According to the conclusions of the analysis,the user portraits of the online learning users are made,and the same points and different points of each group are explored,based on the conventional group division.Creatively,the characteristics of the KOL group belonging to the field are proposed in the field of online education.Based on the characteristics of these four groups,the platform and teachers are proposed strategies and suggestions for curriculum improvement.Finally,the logistic regression algorithm is used to predict the skipping behavior of MOOC learners,and a high-precision predictive model is established.At the same time,the reasons that affect the students'skipping behavior are explored,which provides a reference for the future development of the platform.