AUSWAHL documentation¶

A scikit-learn compatible package for wavelength selection.

Introduction¶

AUSWAHL (AUtomatic Selection of WAvelengtH Library) is a scikit-learn compatible python package that provides a collection of feature selection methods that are intended for spectral datasets.

The goal of AUSWAHL is to provide a unified implementation of several feature selection methods that are popular in the field of chemometrics. Many methods exist (often designed for near-infrared spectra) but comparing them is difficult without a shared code basis.

Using scikit-learn as the underlying framework, allows to integrate the methods from AUSWAHL in your existing machine learning projects. Therewith, all methods known from scikit-learn (e.g. pipelines, cross-validation, hyperparameter search) can also be used with the feature selection methods from the AUSWAHL package. The feature selection methods can be executed by simply calling selector.fit(X, y). Afterwards, the selections are retrieved by calling selector.get_support().

Another goal of AUSWAHL is to provide a large set of feature selection methods as a benchmark for researchers that develop novel feature selection methods.

If you use AUSWAHL, please cite our work:

@manual{auswahl,
    author = {Florian Buckermann and Stefan Heil and Anna Krause},
    title = {AUSWAHL - A scikit-learn compatible package for wavelength selection},
    month = mar,
    year = 2022
}

User Guide

Documentation

Examples

Gallery