# Model Card — Housing Price Prediction Model

## Model Description

This project uses a machine learning regression model to predict housing prices using structured real-estate data. The current implementation uses a Linear Regression algorithm implemented with Scikit-learn. The model is trained using numerical property features such as property size, total number of rooms, and construction year.

The objective of the model is to support data-driven estimation of housing prices within European cities. The implementation is designed primarily for educational and FAIR data science purposes.

---

## Intended Use

The model is intended for educational experimentation, reproducible machine learning workflows, and FAIR data science demonstrations. It may be used to study the relationship between housing characteristics and estimated property prices.

The system is not intended for production-grade real-estate valuation or financial decision-making. Predictions generated by the model should not be interpreted as legally or financially authoritative.

---

## Out-of-Scope Uses

This model must not be used for real-world financial investment decisions, mortgage approvals, insurance risk estimation, or legal property valuation processes.

The model is also not suitable for high-risk automated decision-making scenarios because the training dataset is simplified and limited in scope.

---

## Training Data

The model is trained using structured housing data containing:

- Property size
- Number of rooms
- Construction year
- Listing price

The training workflow includes preprocessing, duplicate removal, missing-value handling, and numeric validation.

---

## Evaluation Results

| Metric | Description |
|---|---|
| RMSE | Root Mean Squared Error |
| MAE | Mean Absolute Error |
| R² Score | Coefficient of Determination |

The current implementation demonstrates a reproducible baseline workflow rather than optimized predictive performance.

---

## Limitations

The current dataset is relatively small and simplified for educational use. Important variables such as neighborhood quality, economic indicators, energy efficiency, and geographic coordinates are not included.

The model therefore cannot capture the full complexity of real-world housing markets.

---

## Ethical Considerations

Machine learning systems for housing prediction may unintentionally introduce socioeconomic biases if trained on unbalanced or incomplete datasets. Historical pricing information may also reflect existing inequalities in housing markets.

This project is intended strictly for FAIR data science education and research demonstration purposes.

---

## License

This model and associated project files are distributed under the MIT License.