A lightweight, plug‑and‑play benchmark kit for synthetic data. Compare Syncora against other generators (e.g., Gretel, MostlyAI) by dropping in CSVs, then auto‑compute fidelity and similarity metrics. Works with any dataset via simple file naming no heavy setup needed.

syncora-benchmarks

Syncora Benchmarks is a plug and play toolkit to evaluate synthetic data quality. Add CSVs from any generator and get instant fidelity metrics and visual comparisons with simple file naming.

What’s Inside

Syncora_vs_Gretel_vs_MostlyAI_metrics_comparison.ipynb
A Jupyter Notebook that:
1. Loads your real & synthetic datasets
2. Computes a suite of similarity & fidelity metrics
3. Visualizes comparative results
README.md
This overview file.
Raw / Synthetic Data Files
Place your CSVs here following the naming convention: _synthetic.csv

Template Usage

Generate synthetic data
Use any platform or in-house model to produce a CSV.
Name your output
Rename your file to: mygenerator_synthetic.csv

e.g. Syncora_synthetic.csv, Gretel_synthetic.csv, etc.

Drop it into this repo
Place your CSV alongside the notebook in the same folder.
Edit & run the notebook

Open Syncora_vs_Gretel_vs_MostlyAI_metrics_comparison.ipynb.
The code automatically discovers all *_synthetic.csv files.
Execute all cells to regenerate metrics and plots.

Adding New Generators or Datasets

Generate your synthetic CSV and name it _synthetic.csv.
(Optionally) Add a short description in the notebook’s metadata.
Re-run the notebook — your new results will be appended to the comparison charts.

Contributing

Fork this repository.
Create a feature branch (git checkout -b feature/xyz).
Submit a pull request with your updates.
Ensure the notebook runs end-to-end without errors.

License

This project is released under the MIT License.