💲 Bitcoin Price Prediction with Deep Learning & Streamlit Web App

This project implements three different deep learning models (CNN, LSTM, GRU) to predict Bitcoin prices using historical OHLCV data. The project includes both a comprehensive training pipeline and an interactive Streamlit web application for real-time price predictions.

🌟 Features

• 🚀 Interactive Web App: User-friendly Streamlit interface for real-time predictions
• 🧠 Multiple Model Architectures: CNN, LSTM, GRU models
• ⚡ Real-time Predictions: Input OHLCV data and get instant price predictions
• 📊 Model Comparison: Compare predictions from all three models
• 🔧 Modular Design: Easy to extend and modify individual components
• 📈 Comprehensive Training Pipeline: Complete data preprocessing and model training

This project was collaboratively developed by:

• Mahmoud Emad Khairy
• Ammar Ahmed Farag
• Mostafa Mohamed Abdullah

📧 Contact & Support

• 🐙 GitHub: mahmoud554377
• 🐙 GitHub: ammarelbordeny
• 🐙 GitHub: HagAli22

📁 Project Structure

bitcoin_price_prediction/
├── data/
│   ├── raw/
│   │   └── BTC_USDT_ohlcv_data.parquet
│   └── processed/
│       └── BTC_USDT_ohlcv_data.csv
├── models/
│   ├── cnn/
│   │   ├── cnn_model.py           # CNN model architecture
│   │   ├── train_cnn.py          # CNN training script
│   │   ├── predict_cnn.py        # CNN prediction script
│   │   └── saved_models/         # Saved CNN models
│   ├── lstm/
│   │   ├── lstm_model.py         # LSTM model architecture
│   │   ├── train_lstm.py         # LSTM training script
│   │   ├── predict_lstm.py       # LSTM prediction script
│   │   └── saved_models/         # Saved LSTM models
│   ├── gru/
│   │   ├── gru_model.py          # GRU model architecture
│   │   ├── train_gru.py          # GRU training script
│   │   ├── predict_gru.py        # GRU prediction script
│   │   └── saved_models/         # Saved GRU models
│   └── ensemble/
│       ├── ensemble_model.py     # Ensemble model architecture
│       ├── train_ensemble.py     # Ensemble training script
│       └── predict_ensemble.py   # Ensemble prediction script
├── utils/
│   ├── __init__.py
│   ├── data_preprocessing.py     # Data preprocessing utilities
│   ├── visualization.py         # Visualization utilities
│   └── evaluation.py            # Model evaluation utilities
├── config/
│   └── config.py                # Configuration parameters
├── notebooks/
│   └── exploration.ipynb        # Original exploration notebook
├── app.py                       # 🌟 Streamlit Web Application
├── scaler.pkl                  # Saved data scaler
├── requirements.txt            # Python dependencies
└── README.md                   # This file

Models Implemented & Performance

📊 Model Comparison Table

Model	Test R²	Validation R²	Test Loss	Val Loss	Notes
🧠 CNN	0.936	0.693	0.002	0.001	Strong on patterns, overfitting signs
📈 LSTM	0.910	0.870	0.003	0.0002	Best generalization
📉 GRU	0.862	0.737	0.005	0.0005	Lightweight & reliable

🚀 Quick Start - Web Application

1. Install Dependencies

pip install -r requirements.txt

2. Run the Streamlit App

streamlit run app.py

3. Use the Web Interface

📱 Web Application Features

1. Open your browser to http://localhost:8501
2. Input Bitcoin OHLCV data:
   • Date: Select prediction date
   • Open Price: Opening price
   • High Price: Highest price
   • Low Price: Lowest price
   • Volume: Trading volume
3. Click "Predict" to get price predictions from all models
4. View results from CNN, LSTM, and GRU models

📋 Installation & Setup

Prerequisites

• Python 3.8 or higher
• pip package manager

Installation Steps

1. Clone the repository:

git clone 
cd bitcoin_price_prediction

2. Create a virtual environment:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. Ensure you have the trained models:
   • btc_cnn_model.pkl
   • btc_lstm_model.h5
   • btc_gru_model.h5
   • scaler.pkl

📈 Data Analysis & Visualizations

Bitcoin Trading Volume Over Time

The trading volume analysis reveals:

• Peak Activity: Massive volume spikes during 2023 (up to 40,000+ units)
• Market Cycles: Volume correlates with major price movements
• Volatility Indicator: High volume periods coincide with market uncertainty

Bitcoin Price Evolution (2018-2025)

Key price milestones:

• 🚀 2021 Bull Run: Reached ~$65,000
• 📉 2022-2023 Correction: Market consolidation around $15,000-$30,000
• 🌙 2025 Surge: Dramatic rise to over $100,000

🎯 Usage Options

Option 1: Web Application (Recommended)

streamlit run app.py

Option 2: Individual Model Training

Train specific models:

# Train CNN model
cd models/cnn
python train_cnn.py

# Train LSTM model
cd models/lstm
python train_lstm.py

# Train GRU model
cd models/gru
python train_gru.py

Option 3: Command Line Predictions

Make predictions with trained models:

# CNN predictions
cd models/cnn
python predict_cnn.py

# LSTM predictions
cd models/lstm
python predict_lstm.py

# GRU predictions
cd models/gru
python predict_gru.py

Configuration

Modify config/config.py to adjust:

• Window size for time series sequences
• Model hyperparameters
• Training parameters
• File paths

Data Requirements

The project expects Bitcoin OHLCV data with the following columns:

• timestamp: Date/time information
• open: Opening price
• high: Highest price
• low: Lowest price
• close: Closing price (target variable)
• volume: Trading volume

Model Architectures

CNN Model

• Two 1D convolutional layers with max pooling
• Dropout layers for regularization
• Dense layers for final prediction

LSTM Model

• Two LSTM layers with return sequences
• Dropout layers for regularization
• Dense layers for final prediction

GRU Model

• Two GRU layers with return sequences
• Dropout layers for regularization
• Dense layers for final prediction

Ensemble Model

• Voting regressor combining LSTM and GRU
• Provides uncertainty estimation through model disagreement

Evaluation Metrics

• R² Score: Coefficient of determination
• MSE: Mean Squared Error
• MAE: Mean Absolute Error
• RMSE: Root Mean Squared Error

Visualization

The project includes comprehensive visualizations:

• Price and volume time series plots
• Training history plots
• Prediction vs actual value plots
• Model comparison plots

Dependencies

Key dependencies include:

• TensorFlow/Keras for deep learning
• scikit-learn for preprocessing and ensemble methods
• pandas for data manipulation
• matplotlib/seaborn for visualization
• joblib for model persistence

License

This project is provided as-is for educational and research purposes.

Contributing

Feel free to open issues or submit pull requests for improvements.

---

requirements.txt

tensorflow>=2.10.0 scikit-learn>=1.1.0 pandas>=1.5.0 numpy>=1.21.0 matplotlib>=3.5.0 seaborn>=0.11.0 joblib>=1.1.0 scikeras>=0.10.0

---