Predicting Academic Achievement Through Engagement Analysis with Sparse Logistic Regression

This research investigates the application of sparse logistic regression for predicting academic success, utilizing engagement data from 410 students across 85,280 learning sessions. Sparse logistic regression, which effectively manages large, feature-rich datasets by eliminating non-essential variables, proved superior to traditional logistic regression in this context. Our analysis focused on three core metrics of student engagement: interaction, feedback, and participation. The study's findings indicate a robust positive correlation between both student interaction and participation with academic achievement, underscoring the critical importance of active engagement in educational success. In contrast, feedback did not significantly impact academic performance, suggesting its influence might be conditional or less directly related to measurable academic outcomes. The advantage of sparse logistic regression in our study highlights its potential as an effective tool for educational data analysis, particularly in environments burdened by high-dimensional data. These results advocate for educational strategies that prioritize interactive and participatory learning experiences over traditional feedback-focused approaches.