Katabatic is an open-source tabular data generation framework designed for data generative models such as GANBLR, TableGAN, MedGan etc...
Katabatic is a framework designed to make generating and evaluating synthetic data much easier. Katabatic has been build with the understanding that different domains have different requirements of synthetic data, and therefore provides a range of evaluation methods.
-
Install Dependencies
-
Download/clone the Katabatic code
The first step is to import the katabatic library:
import katabatic
Next, import the desired model or models:
from katabatic import GANBLR
GANBLR is a data generative model introduced in the following paper:
Zhang, Yishuo & Zaidi, Nayyar & Zhou, Jiahui & li, Gang. (2021).
GANBLR: A Tabular Data Generation Model. 10.1109/ICDM51629.2021.00103.
- Improve the evaluation methods so they are easier to use.
- Debug evaluate_data()
- Move individual models into docker containers.
- Finalise documentation page "Getting Started with Katabatic"
- Finalise documentation page "Installation Guide"
- Cleanup folder structure in preparation to add Katabatic to PyPi.
- Move Aiko implementation from Prototype to Katabatic