Modeling micro-interactions in self-regulated learning: A data-driven methodology

•We developed a data-driven methodology to isolate micro-interactions as indicators of learning progress and achievement.•82% of student achievement variation explained, yielding 91–95% accuracy in identifying low-achieving students.•The added value of non-engagement metrics added a 10% of explainability of student achievement.•These discoveries reveal associations between platform learnability and learning outcome.