Version 2.0.0 Release Notes with ForestDiffusion

This release introduces a generator called ForestDiffusion from the paper "Generating and Imputing Tabular Data via Diffusion and Flow-based XGBoost Models".

Installation: pip install tabgan
Data generation: ForestDiffusionGenerator().generate_data_pipe(train, target, test, )

1.2.0

Full Changelog: Diyago/GAN-for-tabular-data@1.0.1...1.2.0

Updates:

1. Added Timeseries data generation aka TimeGAN
2. Fixed version requirements for colab usage. #22 #23 #24
3. More robust generated data #12
4. Fixed big memory consumption #14
5. Added logs #15

Install: pip install tabgan or pip install tabgan==1.2.0

Release to PIP

https://pypi.org/project/tabgan/1.0.3/

research

move experiment to research folder