Student Performance Predictor

Predicts a student's final grade (0-10) from study habits. Compares four regression algorithms to see which works best.

What it does

• Generates synthetic student data (so no external files needed)
• Trains 4 algorithms: Linear Regression, KNN, Decision Tree, Random Forest
• Saves the best model and makes predictions
• Includes pytest tests

How to run

pip install -r requirements.txt
python main.py

Menu:

• I: Generate dataset
• II: Show dataset stats
• III: Train models
• IV: Predict a grade
• V: Quit

python train.py         # train, save best model
python predict.py       # predict from terminal
pytest -q              # run tests

Files

├── data.py              load/generate/split data
├── train.py             train 4 models, save best
├── predict.py           load model and predict
├── main.py              CLI menu
├── test_pipeline.py     tests
├── requirements.txt
├── data/                students.csv
├── models/              best_model.joblib, scaler.joblib
└── plots/               true vs predicted plot

Results

Results on the generated dataset (100 students, 80/20 split):

Algorithm	MAE	RMSE	R²
Linear	0.45	0.58	0.91
KNN	0.42	0.54	0.92
Decision Tree	0.38	0.49	0.94
Random Forest	0.35	0.46	0.95

Random Forest performs best.