Skip to content

Introduction

AutonFeat is a automatic featurization library for time-series data. It is designed to be used in conjunction with machine learning libraries such as scikit-learn and PyTorch. It is built on design principles similar to that of PyTorch offering a simple and flexible API.

More about the package can be found in here. If you are interested in contributing to the project, please see the contributing guide. If you enjoy using AutonFeat, please consider starring the repository ⭐️.

Why AutonFeat?

What sets this package apart from the many packages already available to researchers and practitioners? We believe that AutonFeat is unique in the following ways:

  • Automatic: AutonFeat is designed to be used with minimal user input. The user only needs to specify the input data and the featurization method(s). The package will automatically featurize the data and return a set of features that can be used for a downstream task.

  • Simple & Interpretable: AutonFeat is designed to be interpretable. The user can understand the featurization process and the resulting features through good documentation.

  • Flexible & Extensible: AutonFeat is designed to be flexible and extensible. The user can easily extend the package to include custom featurization functions.

  • Fast: AutonFeat is designed to be fast enough to be used in production. Our benchmarks prove the utility of our design choices against existing implementations and packages with truly multi-threaded support. Operations are vectorized where possible and parallelized where necessary. We utilize numba and numpy to speed up the featurization process, escaping Python's Global Interpreter Lock (GIL).

Jump In

To get started with AutonFeat, we recommend reading the quickstart guide and following the tutorials we've provided here. If you are interested in the API, you can find the documentation here.

If you enjoy using AutonFeat, please consider starring the repository ⭐️.