How tolgatopaloglu/turkiye_konut_price_predictor works
This is a Turkish housing-price prediction project with a small FastAPI web interface and a documented machine-learning workflow. The serving path is clear in main.py: the app renders a page, accepts housing features at /predict, and calls a loaded model to return a price range. The README describes data preparation, regression experiments, tuning, evaluation and MLflow tracking, but the notebook was not sampled. There is also a filename mismatch: the app opens TOKI_price_regression.pkl1, while the tree contains TOKI_price_regression.pkl; the artifact link is therefore uncertain.
Main entry points: Turkey_Housing_Price_Prediction.ipynb Training workflow · main.py FastAPI app · turkiye_konut_projeleri_2026.csv Housing dataset · index.html Housing page
Components 12
Web serving
- Housing page
templates/index.html - FastAPI app
main.py - Feature input
main.py - Price prediction
main.py
Model development
- Training workflow
Turkey_Housing_Price_Prediction.ipynb - Data preparation
Turkey_Housing_Price_Prediction.ipynb - Model experiments
Turkey_Housing_Price_Prediction.ipynb - Model evaluation
Turkey_Housing_Price_Prediction.ipynb - Experiment tracking
Turkey_Housing_Price_Prediction.ipynb - Serialized model
TOKI_price_regression.pkl
Project data
- Housing dataset
turkiye_konut_projeleri_2026.csv
Other components
- User
Connections 11
- User → FastAPI app: opens page
- FastAPI app → Housing page: renders template
- User → Feature input: submits features
- Feature input → Price prediction: provides input
- FastAPI app → Price prediction: calls predict
- Housing dataset → Data preparation: supplies data
- Training workflow → Data preparation: documents workflow
- Training workflow → Model experiments: documents workflow
- Training workflow → Model evaluation: documents workflow
- Training workflow → Experiment tracking: documents workflow
- Price prediction → Serialized model: loads model