jsyoon0823/TimeGAN

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

1060 stars · 326 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Reference implementation of TimeGAN, a generative adversarial network framework for generating synthetic time-series data, published at NeurIPS 2019. It includes training pipelines, evaluation metrics (discriminative, predictive, PCA/t-SNE visualization), and tutorial notebooks for sine, stock, and energy datasets.

Use cases

• generate synthetic time series data
• augment small time-series datasets for training
• evaluate synthetic time series quality
• reproduce TimeGAN paper results
• benchmark GAN models on sequential data
• create fake stock or sensor data for testing

When to choose

• you need synthetic sequential data for privacy or augmentation
• you want a research baseline for time-series GANs
• you want to reproduce the NeurIPS 2019 TimeGAN results

When to avoid

• you need production-grade, well-supported tooling
• you need non-time-series tabular data generation
• you require a permissive license (license is non-standard)

Facets

library · maturity maintenance

machine-learning deep-learning data-generation machine-learning time-series artificial-intelligence data-science python timegan gan synthetic-data time-series-generation research-code neurips-2019

Member repositories

For agents

markdown · JSON · MCP: product_card(name="jsyoon0823/TimeGAN")