Resample-Lab/README.md at main · mr-september/Resample-Lab · GitHub
Skip to content

README.md

Latest commit

 

History

History
169 lines (118 loc) · 7.16 KB

README.md

File metadata and controls

169 lines (118 loc) · 7.16 KB
Raw

🧪 Resample Lab

Resample Lab provides general guidelines on how to treat imbalanced datasets through a simple interactive interface. Get research-backed recommendations for resampling strategies based on your dataset characteristics.

GitHub release License Live Demo

🔬 Research-Backed | 📊 Interactive | 🎯 Actionable Recommendations

🌐 Try it Live

💖 Support FOSS Projects

My work developing, contributing to, and maintaining open-source software is made possible solely by your donations. Your support is vital to the ongoing development of FOSS solutions.

PayPal Ko-fi Liberapay Crypto donation button by NOWPayments

🌟 Other Ways to Help

⭐ Star the repository to show your support
🐦 Share to help others discover Resample Lab
📝 Write reviews and share your experience
🎥 Create content - tutorials, guides, or showcase videos

📖 Table of Contents

🎯 Features

  • 📊 Interactive Interface - Simple, intuitive web-based tool for dataset analysis
  • 🔬 Research-Backed Recommendations - All suggestions grounded in peer-reviewed literature
  • ⚠️ Regime Detection - Automatic identification of critical dataset characteristics
  • 📈 Strategy Comparison - Visual comparison of different resampling approaches
  • 🎓 Educational - Learn the reasoning behind each recommendation

📚 Citation Sources

The recommendations in Resample Lab are based on peer-reviewed research. Below is a summary of the key sources informing the heuristics.

Citation Year Title Venue Key Contribution
Chawla et al. 2002 SMOTE: Synthetic Minority Over-sampling Technique JAIR, 16, 321–357 Foundational SMOTE algorithm
Drummond & Holte 2003 C4.5, Class Imbalance and Cost Sensitivity: Why Under-Sampling beats Over-Sampling ICML 2003 Undersampling efficiency for large datasets
Batista et al. 2004 A study of the behavior of several methods for balancing machine learning training data ACM SIGKDD Explorations, 6(1), 20–29 SMOTE-Tomek and SMOTE-ENN hybrid methods
Van Hulse et al. 2007 Experimental perspectives on learning from imbalanced data ICML 2007, 935–942 Large-scale resampling benchmarks
He & Garcia 2009 Learning from Imbalanced Data IEEE TKDE, 21(9), 1263–1284 Comprehensive imbalanced learning survey
Blagus & Lusa 2013 SMOTE for high-dimensional class-imbalanced data BMC Bioinformatics, 14(1), 106 SMOTE degradation in high dimensions
Barua et al. 2014 MWMOTE—Majority Weighted Minority Oversampling IEEE TKDE, 26(2), 405–425 Weighted oversampling for hard-to-learn instances
Dal Pozzolo et al. 2015 Calibrating Probability with Undersampling IEEE SSCI 2015 Resampling distorts probability estimates
García et al. 2020 Understanding the apparent superiority of over-sampling Expert Systems with Applications, 158, 113026 Oversampling increases safe minority samples
Fujiwara et al. 2020 Over- and Under-sampling Approach for EISM Data Frontiers in Public Health, 8, 178 HUSDOS-Boost; stratified bagging for EISM
Yang et al. 2024 A review on over-sampling techniques for multi-class imbalanced datasets Frontiers in Digital Health, 6, 1430245 Multi-class oversampling review
Zhao et al. 2025 A Survey on Small Sample Imbalance Problem arXiv:2504.14800 EISM definitions; hybrid ensemble recommendations

🔖 Citation

If you use Resample Lab in your research or project, please cite it as follows:

BibTeX

@software{ResampleLab2025,
  author = {mr-september},
  title = {Resample Lab: Advanced Data Resampling Heuristics},
  year = {2025},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/mr-september/Resample-Lab}}
}

APA

mr-september. (2025). Resample Lab: Advanced Data Resampling Heuristics [Computer software]. GitHub. https://github.com/mr-september/Resample-Lab

⚠️ Critical Regime Thresholds

Based on the literature, Resample Lab identifies three critical regimes:

Regime Trigger Condition Rationale Primary Sources
🔴 Tiny Minority / EISM Minority < 50 samples OR EPV < 10 Synthetic methods may generate unreliable samples; overfitting risk high Zhao et al. 2025, Fujiwara et al. 2020
🟠 High-Dimensional Features > 100 SMOTE fails in high-dimensional spaces due to sparse neighborhoods Blagus & Lusa 2013
🟡 Large-Scale Total samples > 50,000 Undersampling preferred for computational efficiency Drummond & Holte 2003, Van Hulse et al. 2007

🛠️ Getting Started

Quick Start

Visit the Live Demo to start using Resample Lab immediately.

Local Development

# Clone the repository
git clone https://github.com/mr-september/Resample-Lab.git
cd Resample-Lab

# Install dependencies
npm install

# Start development server
npm run dev

🤝 Contributing

Contributions are welcome! Please feel free to submit issues, feature requests, or pull requests.

📄 License

MIT License - see LICENSE file for details.

Star History