Synthesizers

The SDV offers a variety of synthesizers, which use different algorithms to generate synthetic data.

Don't know where to start? We recommend the Fast ML Preset. This is a preset synthesizer that you can use with minimal customization. It's designed to get you going quickly with a basic machine learning model.

Gaussian Copula Synthesizer

Use a classical ML algorithm to learn from real data and generate synthetic data. This synthesizer is the most customizable, with faster performance than other approaches.

CTGAN Synthesizer

Use GAN-based ML algorithm to learn from real data and generate synthetic data. This model can create synthetic data at a high fidelity given enough training time.

＊Day Z Synthesizer

Generate synthetic data from scratch using only the metadata. This synthesizer produces unlimited single table data with the correct formatting.

TVAE Synthesizer

Use a variational autoencoder ML model to learn from real data and generate synthetic data. This model can create synthetic data at a high fidelity given enough training time.

[Experimental] Copula GAN Synthesizer

Use an hybrid ML model to learn from the real data and generate synthetic data. This algorithm combines classicial statistics with GAN-based modeling.

Which synthesizer should I use?

Use the table below to determine the right synthesizer for your needs.

Feature	＊ Day Z	Gaussian Copula	CTGAN	TVAE
Synthesize with only metadata	✔️
Add constraints		✔️	✔️	✔️
Understand & modify		✔️
Anonymize columns	✔️	✔️	✔️	✔️
Learn		✔️	✔️	✔️
Sample with conditions		✔️	✔️(slow)	✔️ (slow)
	✔️	✔️
Availability	Enterprise	Public	Public	Public

＊SDV Enterprise Features. These features are only available for licensed, enterprise users. To learn more about the SDV Enterprise features and purchasing a license, visit our website.