This repo outlines the implementation phase of the "Human Digital Twins - Understanding University Student Behaviour Over Time" research project. This phase focuses on developing predictive (time series) models for student academic performance using clickstream data (Dataset: OULAD) using an LSTM Neural Network, laying the foundation for future extensions like detecting academic exhaustion and procrastination.

This research will compare different baseine models such as a normal LSTM (week vs month predictions), an ANN-LSTM (daily predictions), a CNN-LSTM (week vs month predictions) and a Bidirectional LSTM.

Supervised by Prof Marijke Coetzee