Which CS1 Students Will Fail? Identifying Digital Markers from Learning Analytics in Computer Systems and Architecture Using Weighted Academic Momentum and Interaction Logs

Research from a sub-Saharan African university explores using digital and traditional markers to predict student failure in first-year Computer Systems and Architecture courses. The study, involving four cohorts (2017-2021), aims to identify at-risk students early by analyzing demographic data, self-reported surveys, learning platform interaction logs, and continuous assessment scores, with the goal of enabling timely interventions.

Why it matters

This research is strategically important for institutions focused on educational outcomes and student retention. Identifying at-risk students early can improve educational efficacy, optimize resource allocation for support services, and contribute to workforce development in critical technical fields by ensuring more students successfully complete their foundational studies.

Key insights

• Digital learning platforms generate extensive behavioral data, offering 'digital markers' to identify students who may be struggling.
• The research investigates the predictive power of combining traditional academic data with digital markers for early identification of students at risk of failing.
• The study focuses on a first-year Computer Systems and Architecture (CS1) course at a large public university.
• Data from four cohorts (2017-2021, N=284) were utilized, encompassing demographics, self-reported surveys, Moodle interaction logs, and continuous assessment scores.
• Ten candidate factors were identified through a mixed-methods stakeholder elicitation process.
• A systematic ablation study using logistic regression was employed to analyze the comprehensive feature set.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2608.16914