This project uses machine learning to predict Bitcoin prices across different time horizons (1-day, 7-day, and 30-day forecasts) using scikit-learn's Random Forest Regressor. The model incorporates various technical indicators and market features to make its predictions.

python >= 3.6
yfinance
pandas
numpy
scikit-learn
matplotlib

1. Clone this repository:

git clone <repository-url>

2. Install required packages:

pip install yfinance pandas numpy scikit-learn matplotlib

Simply run the main script:

python bitcoin_predictor.py

The script will:

1. Download Bitcoin historical data from Yahoo Finance (2019-present)
2. Generate technical indicators and features
3. Train models for different prediction windows
4. Create visualization plots saved as PNG files
5. Display performance metrics for each model

• Multiple Moving Averages (SMA & EMA)
• RSI (Relative Strength Index)
• Bollinger Bands
• Volatility Indicators
• Volume Analysis
• Price Momentum
• Daily Returns
• Log Returns

The script generates three visualization files:

• bitcoin_analysis_1day.png: 1-day forecast analysis
• bitcoin_analysis_7day.png: 7-day forecast analysis
• bitcoin_analysis_30day.png: 30-day forecast analysis

Each visualization includes:

• Predicted vs Actual price scatter plot
• Prediction error distribution
• Price trends over time

The model's performance is evaluated using:

• Root Mean Square Error (RMSE)
• Mean Absolute Error (MAE)

• Algorithm: Random Forest Regressor
• Training/Test Split: 80/20
• Features are standardized using StandardScaler
• Model Parameters:
  • n_estimators: 100
  • max_depth: 10
  • random_state: 42

• Past performance doesn't guarantee future results
• Market conditions and external factors can impact accuracy
• Predictions should not be used as sole financial advice

Feel free to fork this repository and submit pull requests for improvements.

This project is licensed under the MIT License - see the LICENSE file for details