Event Log Analysis of In-Class Assignments Using Jupyter Notebook and the Moodle Quiz Module

Abstract

Although CoursewareHub enables Jupyter Notebook log acquisition, established analysis methods are lacking. This study classifies students by treating their behavioral data as event sequences. Using Levenshtein, Normalized Levenshtein, and DTW distances for hierarchical clustering, we analyzed correlations with Quiz scores of Moodle. Findings indicate that high-performing students exhibit specific patterns: approximately 40 events with repeated cell executions, longer deliberation times, and immediate LMS access at the lecture's start.