About EduMind AI
An early-warning risk engine for student academic outcomes.
What it does
EduMind AI predicts a student's end-of-semester CGPA from behavioral, academic, engagement, financial and wellbeing signals, then maps that prediction to an interpretable risk band so mentors can intervene early.
The model
| Type | scikit-learn MLPRegressor (ANN) |
|---|---|
| Architecture | 128-64-32-1 hidden layers (relu activation) |
| Encoded input features | 47 |
| Target | final_cgpa (regression, 0.00–4.00) |
| Pipeline | Impute → scale → one-hot encode → MLPRegressor |
| scikit-learn | 1.6.1 |
Held-out performance (test set)
| MAE | 0.3059 CGPA points |
|---|---|
| RMSE | 0.411 |
| R² | 0.7161 |
| Naive baseline MAE | 0.6479 (predict the mean) |
Risk bands
| Predicted CGPA | Band |
|---|---|
| < 2.50 | High Risk |
| 2.50 – 2.99 | At Risk |
| 3.00 – 3.49 | On Track |
| ≥ 3.50 | Excellent |
Predictions are decision-support estimates, not deterministic outcomes. DIU AI Project Competition 2026 · Team EduMind.