Skip to content

Latest commit

 

History

History
137 lines (91 loc) · 3.58 KB

README_EN.md

File metadata and controls

137 lines (91 loc) · 3.58 KB
Raw

README[中文][English]

MTS: An Efficient Stock Prediction Method Based on the Improved Mamba Model

Project Introduction

MTS (Multiscale Time Series) is an efficient stock prediction method based on the improved Mamba model. This model integrates convolution, attention mechanisms, and multiscale convolution, and introduces a novel local feature extraction module (CFE) to replace traditional convolution operations. The MTS model performs exceptionally well in stock prediction tasks across multiple industries.

Author

Yuanjian Zhang, Group 13, Beijing University of Posts and Telecommunications Email: [email protected]

示例结果

File Structure

MTS/
│
├── mamba_test.ipynb        # Main test script
├── requirements.txt        # Environment dependencies file
└── README.md               # Project description file

Environment Dependencies

Please install the necessary dependencies before running the code. You can install all dependencies listed in requirements.txt using the following command:

pip install -r requirements.txt

Usage Instructions

  1. Clone the Repository

    git clone https://github.com/userpandawin/MambaCFE.git
cd MambaCFE

  2. Install Dependencies

    Ensure you have Python 3.x installed, then install the project dependencies using:

    pip install -r requirements.txt

  3. Prepare Data

    Ensure your data files are located in the data/ directory. If you do not have data, please download or prepare the appropriate stock data.

  4. Run the MTS Model

    Open and run the mamba_test.ipynb Jupyter Notebook file. This Notebook contains the complete code for model training and testing.

Example Code

In mamba_test.ipynb, you will find the following example code:

# Set parameters
class Args:
    use_cuda = True
    seed = 1
    epochs = 90
    lr = 0.01
    wd = 1e-5
    hidden = 16
    layer = 2
    n_test = 46
    ts_code = '301314'  # Select stock code
    cfe = 'True'  # Whether to use CFE
    
args = Args()
args.cuda = args.use_cuda and torch.cuda.is_available()

Evaluation Metrics

The model's performance will be evaluated using the following metrics:

  • MSE (Mean Squared Error)
  • RMSE (Root Mean Squared Error)
  • MAE (Mean Absolute Error)
  • R² (R-squared)

Project Contribution

We welcome any form of contribution, including but not limited to:

  • Submitting bug reports or feature requests
  • Creating Pull Requests for code improvements
  • Providing optimization suggestions

License

This project is licensed under the MIT License. For more details, please refer to the LICENSE file.

Contact

If you have any questions, please contact us at:

Thank you for your attention and support for the MTS project!

Acknowledgments

Special thanks to everyone who contributed and supported this project.

Environment Dependencies File (requirements.txt)

akshare==1.14.29
matplotlib==3.9.1
numpy==2.0.0
pandas==2.2.2
scikit_learn==1.5.1
statsmodels==0.14.2
tensorflow_gpu==2.10.1
torch==2.0.1+cu118
xgboost==2.1.0

Please follow the instructions above to ensure all steps are completed successfully. If you encounter any issues, feel free to contact us!