NFT Music Price Prediction

This project uses machine learning regression and classification models to predict the NFT_LAST_PRICE of music-based NFTs based on metadata, collection-level, and owner-level attributes. Four development phases were implemented using AdaBoost and Random Forest algorithms.

---

📁 Dataset

The dataset used: Training_dataset_Cleaned_all_phasses_updated.csv

• Includes NFT metadata (e.g., first price, sale count)
• Collection details (e.g., average price, ceiling price)
• Owner-related features (e.g., total purchases, diversity)
• Timestamps processed into hour/day/month

---

🔧 Tools and Technologies

• Languages: Python
• Libraries: pandas, numpy, scikit-learn, matplotlib
• Models: RandomForestRegressor, AdaBoostRegressor, RandomForestClassifier, AdaBoostClassifier
• Normalization: MinMaxScaler
• Encoding: LabelEncoder

---

🧠 Model Phases and Features

Phase 1: NFT Metadata Only

• Features: TOKEN_ID, NFT_FIRST_PRICE, NFT_AVG_PRICE, NFT_SALE_COUNT

Phase 2: NFT + Collection Information

• Adds: CONTRACT_ADDRESS, COLLECTION_AVG_PRICE, COLLECTION_CEILING_PRICE, TOTAL_VOLUME_USD

Phase 3: NFT + Owner Information

• Features include owner purchases, diversity, and time features (SALE_HOUR, SALE_DAY, SALE_MONTH)

Phase 4: Combined All Features

• Merges all features from Phases 1–3

---

📊 Evaluation Metrics

• Regression: R², RMSE, MAE
• Classification: ROC AUC, F1 Score, Precision, Recall

---

✅ Results Summary

All models were evaluated using both train/test splits and 5-fold cross-validation.

📉 Regression Results:

• AdaBoost performed better for early phases.
• Random Forest had stronger performance with more complex data (Phases 3 & 4).
• Visual comparisons using RMSE and R² graphs across all four phases are included.

📈 Classification (Phase 4):

• Used a binary label (high/low price based on median).
• ROC Curve plotted for Random Forest and AdaBoost classifiers.

---

📂 Repository Structure

├── data/
│   └── Training_dataset_Cleaned_all_phasses_updated.csv
├── notebooks/
│   └── regression_phase1_to_4.ipynb
│   └── classification_phase4.ipynb
├── outputs/
│   └── model outputs
│   └── ROC_curve_plot.png
├── README.md

---

📌 How to Run

1. Clone the repository

   git clone https://github.com/Haleema33/NFT_Music_Price_Prediction.git

2. Install dependencies

   pip install -r requirements.txt

3. Run each phase notebook in order under /notebooks

---

📘 Future Improvements

• Use advanced models like XGBoost or LightGBM
• Integrate real-time NFT market feeds
• Deploy as a web-based prediction app

---

👩‍💻 Author

Haleema Malik
Final Year Software Engineering Student, Liverpool John Moores University

---

📜 License

This project is open-source and available under the MIT License.

---

📞 Contact

For questions or collaborations:

• Email: haleemamalik589@gmail.com
• GitHub: @Haleema33