TabEval: A Comprehensive Evaluation Framework for Tabular Synthetic Data Generation

TabEval is a comprehensive Python toolkit for evaluating synthetic tabular data generation methods. It provides a unified interface for benchmarking various synthetic data generation algorithms across multiple evaluation dimensions including density estimation, privacy preservation, machine learning efficacy, and structural fidelity.

🎯 Key Features

• Comprehensive Evaluation Metrics: 50+ evaluation metrics across multiple dimensions
• Flexible Benchmarking: Easy-to-use benchmarking suite with caching and reproducibility features
• Extensible Architecture: Plugin-based design for easy integration of new methods and metrics

🚀 Quick Start

Installation

pip install tabeval

📚 Citation

If you use TabEval in your research, please cite:

@misc{tabeval,
  title = {TabEval: A Comprehensive Evaluation Framework for Tabular Synthetic Data Generation},
  author = {Xiangjian Jiang},
  year = {2025},
  howpublished = {\url{https://github.com/SilenceX12138/TabEval}},
}

🙏 Acknowledgments

TabEval builds upon the work of many researchers and open-source projects in the synthetic data generation community. We thank all contributors and the broader research community for their valuable work. In particular, this project builds upon and extends the excellent work from SDMetrics and Synthcity. We acknowledge their foundational contributions to the synthetic data generation ecosystem and have adapted their framework to focus specifically on comprehensive evaluation methodologies.