Real Estate Valuation - Business Analysis with Logistic Regression

A machine learning project that uses logistic regression to predict real estate sale probability and support business decision-making.

📋 Business Analysis Summary

Dataset and Features

Real Estate Valuation Dataset from UCI Machine Learning Repository contains 414 property records with the following key features:

• House Age: Age of the property in years
• Distance to MRT: Distance to nearest Mass Rapid Transit station (meters)
• Number of Convenience Stores: Count of nearby convenience stores
• Unit Price: House price per unit area (NT$/m²)
• Location: Latitude and longitude coordinates

Classification Problem

Objective: Predict whether a real estate property will sell based on its characteristics and listing price.

Target Variable: Binary classification

• 1 = Property will likely sell (priced within 120% of market average)
• 0 = Property will likely NOT sell (priced above 120% of market average)

Model Performance:

• Training Accuracy: 99.7%
• Test Accuracy: 99.2%

Business Decision Support

This logistic regression model supports several critical business decisions:

1. Pricing Strategy: Determine optimal listing price to maximize sale probability
2. Market Analysis: Identify properties likely to sell quickly vs. those requiring price adjustments
3. Investment Decisions: Assess property attractiveness based on location and characteristics
4. Risk Assessment: Quantify the probability of a successful sale before listing

Business Impact: Real estate agents and property investors can use this model to make data-driven pricing decisions, reducing time on market and maximizing profitability.

🚀 Project Features

• Jupyter Notebook Analysis: Complete data exploration, model training, and evaluation
• Interactive Streamlit App: Web interface for real-time property sale probability predictions
• Model Persistence: Trained model saved for production use
• Visualization: Price vs. probability charts and business insights

📁 Project Structure

real_estate/
├── businessAnalysis_realEstateValuation.ipynb  # Main analysis notebook
├── real_estate_app.py                         # Streamlit web application
├── real_estate_model.pkl                      # Trained logistic regression model
├── data/
│   ├── Real estate valuation data set.xlsx    # Original dataset
│   └── real+estate+valuation+data+set.zip    # Backup data
└── README.md                                  # This file

🛠️ Installation and Setup

1. Clone the repository:

   git clone https://github.com/mobius29er/realEstateValuation.git
   cd realEstateValuation

2. Install required packages:

   pip install pandas numpy scikit-learn matplotlib streamlit joblib openpyxl

3. Run the Jupyter Notebook:

   jupyter notebook businessAnalysis_realEstateValuation.ipynb

4. Launch the Streamlit App:

   streamlit run real_estate_app.py

🌐 Streamlit Cloud Deployment

To deploy this app on Streamlit Cloud:

1. Fork or clone this repository to your GitHub account
2. Visit share.streamlit.io
3. Connect your GitHub account and select this repository
4. Set the main file path to: real_estate_app.py
5. Deploy - Streamlit Cloud will automatically install dependencies from requirements.txt

Live Demo: The app can be deployed and accessed via Streamlit Cloud for interactive demonstrations.

💡 Usage

Jupyter Notebook

Open businessAnalysis_realEstateValuation.ipynb to explore:

• Data loading and preprocessing
• Model training and evaluation
• Price sensitivity analysis
• Business insights and recommendations

Streamlit Web App

Launch the app to interactively:

• Input property characteristics (age, MRT distance, convenience stores, price)
• Get real-time sale probability predictions
• Receive business recommendations
• Visualize results with color-coded feedback

📊 Model Details

• Algorithm: Logistic Regression
• Features: House age, MRT distance, convenience stores count, listing price
• Target: Binary classification (will sell / won't sell)
• Performance: 99.2% test accuracy
• Business Rule: Properties priced within 120% of market average are likely to sell

🎯 Business Applications

This model can be used by:

• Real Estate Agents: Optimize listing prices for faster sales
• Property Investors: Assess investment opportunities
• Homeowners: Determine competitive pricing strategies
• Market Analysts: Understand pricing dynamics

📈 Key Insights

• Properties priced within 120% of market average have high sale probability
• Distance to MRT station significantly impacts sale likelihood
• Number of nearby convenience stores affects property attractiveness
• House age influences buyer preferences

🔧 Technical Stack

• Python 3.13+
• pandas: Data manipulation and analysis
• scikit-learn: Machine learning algorithms
• matplotlib: Data visualization
• streamlit: Web application framework
• joblib: Model serialization

📝 License

This project is part of an educational assignment for AI/ML coursework.

🤝 Contributing

This is an educational project. Feel free to fork and experiment with different models or features.