EchoData — Synthetic Data for Data Protection

A cross-platform web GUI for generating privacy-preserving synthetic data and comparing it against the original using machine learning.

Works on macOS, Windows, and Linux — runs as a local Flask server and opens in your browser.

---

Quick Start

# 1. Install dependencies
pip install -r requirements.txt

# 2. Run (opens browser automatically)
python app.py

Then open http://localhost:5111 if it doesn't open automatically.

How It Works

1. Load — Drag-and-drop any CSV file. Auto-detects numeric and categorical columns.
2. Configure — Adjust the privacy budget (ε) and number of synthetic rows.
3. Generate — The ShatteredSynth engine decomposes your data into noisy statistical fragments and reassembles synthetic rows.
4. Compare — Train Random Forest models on both datasets to verify predictive patterns are preserved.