🏢 24S2-CB0494 Singapore HDB Resale Price Prediction 📈

🔗 Project Links

• GitHub Repo
• View Rendered Notebook

🔭 Overview

This project leverages machine learning algorithms to accurately predict Housing & Development Board (HDB) resale flat prices in Singapore. By applying advanced regression techniques to a comprehensive dataset of over 200,000 transactions, we've created a powerful price prediction model that achieves an R² score exceeding 0.87 on test data.

📊 Critical Performance Metrics

Model	Test RMSE (SGD)	Test R²	Test MAE (SGD)
Linear Regression	37,856.23	0.8523	28,342.15
Decision Tree	34,761.41	0.8749	23,691.32

🔍 Critical Findings

1. Most Important Factors Influencing Resale Prices:
   • Floor area is the strongest predictor, with larger units commanding significantly higher prices.
   • Location (town) has a significant impact, with central regions showing premium pricing patterns.
   • Flat type creates distinct pricing tiers, with 5-room and executive flats at a premium.
   • Floor level demonstrates a consistent positive correlation with price.
   • Remaining lease shows an evident influence on valuation decisions.

2. Model Performance:
   • Feature importance analysis reveals actionable insights for buyers and sellers.
   • Our optimized Decision Tree model captures complex non-linear relationships in the housing market.
   • The model handles categorical variables effectively through sophisticated preprocessing.

3. Practical Applications:
   • Insights provide strategic advantages for both buyers and sellers in negotiation contexts.
   • Interactive pricing tool allows stakeholders to estimate property values with confidence.
   • Methodology establishes a framework for ongoing market analysis as conditions evolve.

💻 Technical Implementation

This project demonstrates mastery of data science techniques, including:

• Advanced feature engineering (lease calculation, categorical encoding, temporal feature extraction).
• Hyperparameter optimization via grid search cross-validation.
• Interpretable visualizations for complex multidimensional data.
• Robust model evaluation with industry-standard metrics (MAE, RMSE, R²).
• Sophisticated data preprocessing (pipelines, scalers, one-hot encoding).

🛠️ Setup and Usage

# Clone this repository  
git clone https://github.com/gracenngg/24S2-CB0494-HDB-Price-Predictions.git  

# Navigate to the repository  
cd hdb-price-predictions  

# Install required packages  
pip install -r requirements.txt  

# Run the Jupyter notebook  
jupyter notebook HDB_Resale_Price_Predictions.ipynb  
=======  
# 24S2-CB0494: HDB Price Predictions  
 HDB Flat Price Predictions Using Machine Learning  
>>>>>>> 93188c1111653fcb7124f46ac0fbe19684a28a28