🏠 Asturias Housing Price Predictor

---

📝 Project Description

This project applies Machine Learning techniques to predict housing prices in Asturias, Spain, using real-world data collected from Idealista.

The dataset includes information such as:

• Property type
• Location (neighborhood, municipality, district)
• Built area (square meters)
• Number of bedrooms and bathrooms
• Selling price

The goal is to develop a tool that helps buyers, real estate agents, or analysts to make informed decisions in the property market.

---

🎯 Objectives

• 🧠 Predict housing prices based on property features.
• 📊 Identify the most influential variables affecting price (square meters, location, etc.).
• ⚖️ Compare different Machine Learning models to choose the most suitable one.
• 🌍 Provide a practical solution for analyzing the real estate market in Asturias.

---

📑 Project Structure

📂 ML_AsturiasHousingPricePredictor/
├── 📂 src/
│   ├── 📂 data_sample/
│   │   ├── cleaned_dataset.csv
│   │   └── scrap_data_asturias.csv
│   ├── 📂 img/
│   │   ├── additional_features.png
│   │   ├── correlation_heatmap_extended.png
│   │   └── ...
│   ├── 📂 models/
│   │   ├── xgb_full_model.pkl
│   │   └── xgb_reduced_model.pkl
│   ├── 📂 notebooks/
│   │   ├── eda.ipynb
│   │   └── preprocessing.ipynb
│   ├── 📂 results_notebook/
│   │   └── final_results.ipynb
│   └── 📂 utils/
│       ├── bootcampviztools.py
│       └── toolbox_ML.py
│
├── .gitignore
├── README.md
└── requirements.txt

---

🛠️ Technologies Used

• Python
• Pandas, NumPy
• Matplotlib, Seaborn
• Scikit-learn
• XGBoost, CatBoost, LightGBM
• Jupyter Notebook

---

🚀 How to Run the Project

1. Clone the repository:

   git clone https://github.com/apvwbm/ML_AsturiasHousingPricePredictor.git
   cd ML_AsturiasHousingPricePredictor

2. (Optional) Create a virtual environment:

   python -m venv venv
   source venv/bin/activate # macOS/Linux
   venv\Scripts\activate # Windows

3. Install the dependencies:

   pip install -r requirements.txt

4. Explore the notebooks:

   jupyter notebook src/notebooks/*.ipynb

5. Check the final results:

   jupyter notebook src/results_notebook/final_results.ipynb

---

📊 Results

The selected model was XGBoost, trained using only the 7 most influential features according to permutation_importance. It achieved the following results:

• MAE: €7,446
• RMSE: €46,321
• R²: 0.9175

📌 This demonstrates that precise and efficient housing price predictions are achievable even with a reduced set of variables.

---

✅ Conclusions

• The model can estimate housing prices in Asturias with an accuracy of 91.7% (R²).
• Built area and neighborhood location are the most determining factors.
• The number of features was successfully reduced without sacrificing accuracy, enhancing efficiency.
• This approach is adaptable to other regions and serves as a basis for automated valuation tools.

---

👤 Author

Aitor Pérez

LinkedIn | GitHub