ZenitAI

Description

ZenitAI is a Python library focused on providing convenient tools for machine learning tasks. For now the library is focused on binary classification tasks.

What's inside

The main component of the library: A tool for conducting WOE-analysis in classification tasks on tabular data.

It is a highly customizable class facilitating data transormation with almost no effort. Class uses standard sklearn-like interface for transformation

Library also includes:

1. some wrapper functionss allowing avoiding a considerable amount of boilerplate needed with working with skikit-learn.
2. a tool for managing multiple launches of machine-learning models (in early development stage)

Installation

pip install zenitai

Documentation

Documentation can be found on Read the Docs https://zenit-ai.readthedocs.io/en/latest/

Quickstart

:TBD:

Examples

:TBD:

Limitations

:TBD:

TODO list

TODO:

• Add to_pickle method for WoeTransformer
• Add error logging
• Add tqdm integration
• Add some usecases for WoeTransformer
• Add option for auto-detection of categorical variables (say, 15 unique values or lower)
• Add handling of unknown categories (don't remember if there's any)
• Add handling of Pandas Categorical type