Bitcoin Price Prediction using CNN + LSTM and Random Forest

This project demonstrates how to predict Bitcoin prices using a combination of CNN + LSTM neural network models and a Random Forest Regressor. The project leverages historical data from the CoinRanking API to train and evaluate models for accurate predictions.

📑 Table of Contents

• 📃 Project Overview
• 📊 Data Source
• 📈 Models Used
  • CNN + LSTM
  • Random Forest Regressor
• ⚙️ Model Training Process
• 🚀 Installation
• 💻 Usage
• 🏆 Results
• 🤝 Contributing

📃 Project Overview

The goal of this project is to predict the closing price of Bitcoin using two different models:

1. A hybrid CNN + LSTM model, which is well-suited for time series data.
2. A Random Forest Regressor, a robust ensemble learning method.

These models are trained using historical Bitcoin price data retrieved from the CoinRanking API over a period of 5 years. Predictions are made using a combination of these models, and their performance is evaluated based on common metrics like MAE, MSE, and MAPE.

📊 Data Source

The data is fetched using the CoinRanking API, which provides historical price data for Bitcoin. The historical data includes:

• Timestamps
• Closing prices

API Details:

• Endpoint: /coin/{coin_id}/history
• Time Period: 5 years
• Coin ID for Bitcoin: Qwsogvtv82FCd

📈 Models Used

CNN + LSTM

This model is particularly effective for sequential data. The architecture includes:

• LSTM Layer: To capture temporal dependencies in the data.
• Dense Layers: To perform the final regression after feature extraction.
• Dropout and Batch Normalization: To prevent overfitting and ensure stable training.
• Weighted MSE Loss: The custom loss function emphasizes recent data points.

Random Forest Regressor

An ensemble method that operates by constructing multiple decision trees during training and outputting the mean prediction of individual trees.

⚙️ Model Training Process

Data Preprocessing

• Scaling: The MinMaxScaler is used to normalize the Bitcoin prices for better training performance.
• Data Splitting: 60% of the data is used for training, and 40% is reserved for testing.
• Sliding Window Approach: A window of 60 days' worth of data is used to predict the next day's closing price.

CNN + LSTM Model

1. Architecture:
   • LSTM layers with 50 units.
   • Dense layers with ReLU activation and regularization.
   • Dropout layers to reduce overfitting.
2. Optimizer: Stochastic Gradient Descent (SGD) with dynamic learning rate and momentum.
3. Loss Function: A custom weighted MSE loss is used to prioritize recent data points.

Random Forest Model

• Parameters:
  • 100 trees.
  • Maximum depth of 10.
  • Minimum samples split of 5.

Cross-Validation

• K-Fold Cross-Validation (with k=10) is employed to ensure the robustness of the Random Forest model.

Evaluation Metrics

The following metrics are used to evaluate the performance of both models:

• MAE (Mean Absolute Error)
• MSE (Mean Squared Error)
• MAPE (Mean Absolute Percentage Error)

🚀 Installation

To run the project locally, follow these steps:

1. Clone the repository:

   git clone https://github.com/yourusername/bitcoin-price-prediction.git
   cd bitcoin-price-prediction

2. Install the required dependencies:

   pip install -r requirements.txt

3. Set up the CoinRanking API: Update the x-access-token in the code with your own API token from CoinRanking.

💻 Usage

1. Run the model training:

   python model_training.py

2. View the plots:

   • The real vs predicted prices will be plotted and saved as PNG files in the project directory.

3. Analyze Results:

   • The CNN + LSTM model and the Random Forest model's performance metrics will be printed in the console.

🏆 Results

CNN + LSTM

• MAE: 0.1460
• MSE: 0.0267
• MAPE: 1.8463
• ACCURACY: 98.15%

Random Forest

• MAE: 0.0935
• MSE: 0.0121
• MAPE: 0.8317
• ACCURACY: 99.16%

The comparison between the two models is visualized in the generated plot that displays real prices, CNN + LSTM predictions, and Random Forest predictions.

🤝 Contributing

Contributions are welcome! Feel free to open an issue or submit a pull request.