Data-Centric-AI-Community/fg-data-synthetic

Features

• Synthetic Data Generators - Creates realistic tabular and time-series datasets that preserve the statistical properties of original sources for machine learning.
• Tabular - Generates privacy-preserving tabular datasets that maintain the statistical properties and data quality of original sources.
• Synthetic Time Series Generation - Produces synthetic sequences that replicate the temporal patterns and characteristics of original data while protecting sensitive information.
• Privacy-Preserving Machine Learning - Creates realistic datasets that maintain original statistical properties while protecting sensitive information during machine learning development.
• Differential Privacy Noise Injection - Ensures privacy by injecting controlled mathematical noise into generated datasets to prevent individual re-identification.
• CPU-Optimized Statistical Sampling - Generates synthetic data using CPU-optimized mathematical operations to ensure high performance on standard hardware.
• Machine Learning Data Augmentation - Expands training datasets with synthetic examples to improve model performance and robustness without requiring additional real-world data collection.
• Data Synthesis - Generates synthetic tables for training machine learning models when original data is restricted by privacy concerns.
• Tabular Data Synthesizers - Produces privacy-compliant tables by modeling the underlying statistical distributions of source data without requiring specialized hardware.
• Statistical Modeling - Learns the underlying probability distributions of input data to sample new records that maintain original statistical properties.
• High-Performance Data Generation - Enables high-performance generation of realistic datasets using statistical models on standard hardware to accelerate testing and development workflows.
• Time Series Generation - Creates synthetic temporal sequences that replicate complex patterns and characteristics found in original time-based datasets.
• Time-Series Data Simulation - Produces realistic temporal sequences that replicate complex patterns and characteristics of original data for testing and predictive modeling.
• Time Series Modeling - Captures sequential correlations and time-based patterns to ensure generated time-series data maintains realistic chronological flow.
• Synthetic Data CLI Automations - Provides a command-line interface for configuring and executing automated workflows to create realistic datasets without manual coding.
• Declarative Workflow Definitions - Executes data generation pipelines through a configuration-driven engine that maps user-defined parameters to specific statistical modeling tasks.
• Data Privacy Protection Tools - Provides a platform that generates synthetic information to protect sensitive records while maintaining data utility for testing and model training.
• Schema-Aware Synthesis - Validates generated records against predefined structural constraints to ensure compatibility with existing database and machine learning schemas.

Star history