House Price Prediction

End-to-end ML pipeline predicting California house prices with strong results.

Results

• RMSE: 0.4392 (original scale)
• MAE: 0.2793
• R²: 0.8528
• Baseline RMSE: 1.1583 → Improvement: 62.1% (target exceeded)

Quickstart

• Install: pip install -r requirements.txt
• Train & evaluate: python src/test_models.py (writes results/final_project_report.txt, saves models in models/)
• Production demo: python src/ml_pipeline.py (runs a sample single prediction)

Project Structure

src/
  data_loader.py        # Load/save dataset
  eda_analysis.py       # EDA and plots
  data_preprocessor.py  # Feature engineering, scaling, outliers, target log
  model_trainer.py      # Model training, CV, tuning, report
  ml_pipeline.py        # Prediction pipeline (load preprocessor/model, predict)
  test_preprocessing.py # Preprocessing sanity check
  test_models.py        # Full training + final report
results/
  final_project_report.txt, plots
models/
  preprocessor.pkl, best_model.pkl, others

Notes

• Best model: XGBoost (top CV RMSE)
• Consistent preprocessing: saved preprocessor ensures identical train/inference transforms

Requirements

• Python 3.8+
• See requirements.txt for exact package versions