DOCSEARCHbyAlgolia
Submit Docs

Synthetic Data Studio Documentation

website·Indexed 31 days ago

https://docs.synthdata.studio

A production-ready, open-source platform for generating high-quality synthetic datasets. Built with Python, FastAPI, PostgreSQL, Redis, Docker, and PyTorch, it features AI-powered insights, differential privacy guarantees (Epsilon and Delta), and comprehensive quality assessment using statistical similarity and ML utility metrics. The documentation covers installation, API interaction via cURL and Python, and deployment options via Docker, Heroku, and Railway.

Tokens
55.2K
Snippets
283
Records
384
Agent score
62%

What's inside Synthetic Data Studio

  1. Overview of Synthetic Data Studio

    Synthetic Data Studio is a production-ready platform designed to generate high-quality synthetic datasets with mathematical privacy guarantees. It allows organizations to create realistic data for development, testing, and analytics without exposing sensitive information, focusing on three core pillars: Privacy First, Quality Assured, and Enterprise Ready (compliance and audit capabilities).

  2. Overview of Synth Studio synthetic data generation

    Synth Studio is a platform designed to generate high-quality synthetic data with mathematical privacy guarantees. It is specifically built for handling sensitive datasets that must comply with regulations such as HIPAA (healthcare), GDPR (financial/general), CCPA, and SOC-2.

  3. Overview of Synth Studio synthetic data platform

    Synth Studio is a privacy-preserving synthetic data generation platform designed to create high-quality synthetic datasets with mathematical privacy guarantees. It is specifically built for handling sensitive datasets subject to regulations such as HIPAA (healthcare), GDPR (financial/general), CCPA, and SOC-2.

  4. Overview of Synth Studio synthetic data platform

    Synth Studio is a privacy-preserving synthetic data generation platform designed to create high-quality synthetic datasets with mathematical privacy guarantees. It is specifically built for handling sensitive data under regulatory frameworks such as HIPAA (healthcare), GDPR (financial/general), and CCPA.

  5. Overview of Synth Studio synthetic data generation

    Synth Studio is a privacy-preserving synthetic data generation platform designed to create high-quality synthetic datasets with mathematical privacy guarantees. It is intended for use with sensitive datasets subject to regulations such as HIPAA (healthcare), GDPR (financial/general), CCPA, and SOC-2.

  6. Overview of Synth Studio technology stack and features

    Synth Studio is a free and open-source (MIT License) platform for generating privacy-preserving synthetic data. It is built using Python, FastAPI, PostgreSQL, Redis, Docker, and PyTorch. Key capabilities include:

    • Data Generation: Supports multiple models including CTGAN, TVAE, GaussianCopula, and TimeGAN with GPU acceleration.
    • Privacy: Implements Differential Privacy with RDP (Rényi Differential Privacy) accounting to prevent re-identification.
    • Evaluation: Provides quality metrics for statistical similarity, ML utility, and privacy leakage detection.
    • Compliance: Includes reporting packs for HIPAA, GDPR, and CCPA with audit logs and model cards.
    • Enterprise Features: Supports asynchronous jobs, PostgreSQL/S3 storage, and Docker-ready deployment.

  7. Overview of Synth Studio capabilities

    Synth Studio is an open-source (MIT License) platform for generating privacy-preserving synthetic data. Key capabilities include:

    • Data Generation: Supports multiple GPU-accelerated generators including CTGAN, TVAE, GaussianCopula, and TimeGAN.
    • Privacy & Compliance: Implements Differential Privacy with RDP accounting to prevent re-identification and provides compliance packs for HIPAA, GDPR, and CCPA including audit logs and model cards.
    • Evaluation: Includes quality metrics for statistical similarity, ML utility, and privacy leakage detection.
    • AI Integration: Features AI-driven insights for chatting with data, smart suggestions, and auto-documentation.
    • Infrastructure: Built for enterprise use with async jobs and support for PostgreSQL and S3 storage.

  8. Synthetic Data Studio Platform Overview

    Synthetic Data Studio is a production-ready platform for generating high-quality synthetic data with differential privacy guarantees. It supports multiple synthesis methods including CTGAN, TVAE, and GaussianCopula, and is built on a scalable architecture using FastAPI and SQLAlchemy with Docker/cloud-native readiness.

  9. Understand Epsilon (ε) in Differential Privacy

    Epsilon (ε) is the privacy budget in differential privacy. It defines the trade-off between privacy protection and data utility: a lower ε provides stronger privacy by adding more noise, while a higher ε provides better utility with weaker privacy protection.

    Privacy Loss Interpretation:

    • < 0.1: Negligible loss (Extremely strong privacy)
    • 0.1 - 1.0: Very Low loss (Excellent privacy)
    • 1.0 - 5.0: Low loss (Strong privacy)
    • 5.0 - 10.0: Moderate loss (Reasonable privacy)
    • 10.0 - 20.0: High loss (Weak privacy)

    • 20.0: Very High loss (Minimal privacy)

  10. Synth Studio Technical Stack Overview

    Synth Studio is a free and open-source synthetic data platform (MIT License) built using the following industry-standard technologies: Python, FastAPI, PostgreSQL, Redis, Docker, and PyTorch.

  11. Developer Setup and Integration Path

    Developers should follow this sequence to integrate with or contribute to the platform:

    1. Configure the development environment.
    2. Review the system architecture.
    3. Explore API examples.
    4. Implement testing guidelines.

  12. Use AI-powered analysis features

    The API provides LLM-powered endpoints to chat about specific evaluations, get improvement suggestions for a dataset, or get an explanation for a specific metric and its value.
    # Chat About Evaluation
curl -X POST "http://localhost:8000/llm/chat" \
  -H "Authorization: Bearer $ACCESS_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "message": "What is the overall quality of this synthetic data?",
    "evaluation_id": "your-evaluation-id",
    "history": []
  }'

# Get Improvement Suggestions
curl -X POST "http://localhost:8000/llm/suggest-improvements/{evaluation_id}" \
  -H "Authorization: Bearer $ACCESS_TOKEN"

# Explain Metric
curl -X GET "http://localhost:8000/llm/explain-metric?metric_name=ks_statistic&metric_value=0.087" \
  -H "Authorization: Bearer $ACCESS_TOKEN"