Analyzing Students' Behavior in Blended Learning Environment for Programming Education

Abstract

Analyzing students' test scores and online learning behaviors in the blended learning environment of programming education can help computer educators understand the students' learning and programming process. Furthermore, it can help teachers provide personalized guidance to students. In this article, we first performed a time-series clustering algorithm on the 5 test scores of students online and offline, and obtained three distinct student types (Excellent, Moderate, Poor). To further study the behavior of students, a correlation analysis of the four behavior data obtained from online programming and test scores was conducted. Students' online time is negatively correlated with the test score. Then, a cluster analysis of the online behaviors data was conducted, and three different student types were also obtained (Poor performance, High-quality learning, Learning hard). Finally, the two clustering results were compared. In the recognition of middle-level students, their similarity was 10/15. Long-term online learning can achieve good and stable test scores. The consistency of student behavior qualitatively proves the rationality of our research. Besides, for students with large differences in the two clustering results, we provided a targeted analysis and gave teachers corresponding suggestions.