Single Table Data

SDV supports modeling single table datasets. It provides unique features for making it easy for the user to learn models and synthesize datasets. Some important features of sdv.tabular include:

• Support for tables with primary key

• Support to anonymize certain fields like addresses, emails, phone numbers, names and other PII information.

• Support for a number of different data types - categorical, numerical, discrete-ordinal and datetimes.

• Support multiple types of statistical and deep learning models:

  • GaussianCopula Model: A tool to model multivariate distributions using copula functions.

  • CTGAN Model: A GAN-based Deep Learning data synthesizer that can generate synthetic tabular data with high fidelity.

• Data Description
  • Table Metadata
• Models
  • Tabular Preset
  • GaussianCopula Model
  • CTGAN Model
  • CopulaGAN Model
  • TVAE Model
• Constraints
  • Handling Constraints
  • Defining Custom Constraints

User Guides Data Description