I’m sharing a synthetic loan default dataset I created and used during the Deep Learning IndabaX Zimbabwe 2026 Hackathon.

The dataset is designed for machine learning practice around credit risk and loan default prediction. It is fully synthetic, built to reflect realistic patterns while keeping data privacy in mind. It can be useful for learning, experimentation, and building baseline models in financial AI.

You can access it here: https://zenodo.org/records/20569260

DOI for citation: 10.5281/zenodo.20569260

For anyone who uses it, please cite the DOI above.

Happy to hear feedback or see what others build with it.