Synthetic Data Generator

This tool generates synthetic datasets from a user-defined schema, supporting numeric, categorical, and boolean features with configurable sample size. It enables reproducible experiments by allowing seed-based generation and the preservation of feature relationships through covariances or copulas. It also offers optional privacy controls to inject noise or apply DP-inspired adjustments. The generated datasets facilitate ML model training, ETL testing, and privacy-safe data sharing without exposing real records.
Conceptually, you define per-feature definitions (name, type, distribution parameters) and optional correlations; the engine draws samples from the specified margins and applies transformations to achieve the target distributions and relationships. Outputs include the dataset plus a detailed generation metadata block for auditability. The tool supports multiple export formats such as CSV, JSON, and Parquet, enabling easy integration with data science notebooks and pipelines.
Who benefits: data scientists, ML engineers, QA testers, and data engineers who need repeatable data, realistic distributions, and controlled privacy. Unique value stems from a schema-driven approach combined with reproducibility, provenance metadata, and privacy-aware options that can be tuned to project requirements. Core features include a schema editor, distribution templates, correlation controls, seed-based reproducibility, and export-ready formats for immediate reuse in modeling or testing scenarios.
Typical use cases include ML model prototyping, pipeline validation, and privacy-conscious data sharing for internal experimentation.