SDV
Synthetic data generation for tabular data
What is SDV?
Synthetic data generation for tabular data
Overview of SDV
What it is
SDV (Synthetic Data Vault) is a Python library for creating tabular synthetic data. It is part of The Synthetic Data Vault Project from DataCebo. The library uses machine learning algorithms to learn patterns from real data and emulate them in synthetic data. It supports generating data for single tables, multiple connected tables, or sequential tables.
Key features
- Create synthetic data using machine learning: offers multiple models ranging from classical statistical methods (GaussianCopula) to deep learning methods (CTGAN).
- Evaluate and visualize data: compare synthetic data to real data against a variety of measures, diagnose problems, and generate a quality report.
- Preprocess, anonymize and define constraints: control data processing to improve synthetic data quality, choose from different anonymization types, and define business rules as logical constraints.
Use cases
SDV is designed for generating synthetic tabular data when real data is sensitive, scarce, or cannot be shared. It can be used to create anonymized datasets for testing, development, or machine learning model training while preserving statistical patterns and relationships between columns. The library also supports evaluating the quality of generated synthetic data against the original real data.
Getting started
Install SDV using pip or conda. It is recommended to use a virtual environment.bashpip install sdv``````bashconda install -c pytorch -c conda-forge sdvAfter installation, load a demo dataset and create a synthesizer:pythonfrom sdv.datasets.demo import download_demoreal_data, metadata = download_demo(modality='single_table', dataset_name='fake_hotel_guests')from sdv.single_table import GaussianCopulaSynthesizersynthesizer = GaussianCopulaSynthesizer(metadata)synthesizer.fit(data=real_data)synthetic_data = synthesizer.sample(num_rows=500)
License and pricing
The SDV is publicly available under the Business Source License. The pricing type is open source.
Based on the project's public README and repository metadata.