Using AI to forecast student dropout risk in technical education using a learning analytics approach.

The integration of artificial intelligence and learning analytics in higher education creates new opportunities for adaptive curriculum design and evidence-based decision-making. This study investigates how predictive analytics can improve e-course development, detect early signs of students at risk of dropping out, and support timely pedagogical interventions. Using data from six CAD course logs delivered via Moodle at TTK University of Applied Sciences (Estonia) during the 2024–2025 academic year (99,104 activity records from 154 students), a machine learning model was developed to identify students at risk of academic failure. The proposed weighted attribute method combines key metrics—engagement, learning difficulty, and time allocation—to standardise behavioural and cognitive factors for outcome prediction. Logistic regression and decision tree classifiers were trained and assessed. The findings show that predictive models can identify at-risk students, allowing early alerts and adaptive support through ongoing monitoring. The framework turns learning analytics into actionable insights for educators, enabling proactive measures. The study concludes that AI-driven analytics, developed within ethical and pedagogically sound frameworks, can enhance transparency and foster equitable learning in higher education.