Installation

pip install dataroid

Example Usage

from dataroid import Bot
import pandas as pd

data = pd.read_csv("shopping.csv")

model = Bot(data)
model.generate(5)

Advanced Usage

from dataroid import Bot
import pandas as pd

data = pd.read_csv("shopping.csv")

params = {
    'epochs':5,
    'embedding_dim':64,
    'generator_dim':(256,256),
    'discriminator_dim':(256,256),
    'generator_lr':0.0002,
    'generator_decay':1e-06,
    'discriminator_lr':0.0002,
    'discriminator_decay':1e-06
}

model = Bot(data,**params,sampling=True)
model.generate(5)

Dependencies

• pandas - A Python package that provides fast, flexible, and expressive data structures designed to make working with "relational" or "labeled" data both easy and intuitive.
• ctgan - CTGAN is a collection of Deep Learning based synthetic data generators for single table data, which are able to learn from real data and generate synthetic data with high fidelity.

Citation

Lei Xu, Maria Skoularidou, Alfredo Cuesta-Infante, Kalyan Veeramachaneni. Modeling Tabular data using Conditional GAN. NeurIPS, 2019.

@inproceedings{ctgan,
  title={Modeling Tabular data using Conditional GAN},
  author={Xu, Lei and Skoularidou, Maria and Cuesta-Infante, Alfredo and Veeramachaneni, Kalyan},
  booktitle={Advances in Neural Information Processing Systems},
  year={2019}
}